NullPointerException when running specjvm08 compiler.compiler with jdk in build snv_101
Comments
EVALUATION
This is a duplicate of 6636352. It turns out the build that introduced the autobox changes also included this fix so it was never reproducible in a promoted build of jdk 7.
10-04-2009
EVALUATION
I built my own version of HS11b09 and it fails with JDK 6 and even with JDK 7b54.
10-04-2009
EVALUATION
6u4p-b03 uses the same HS version and has the same bug:
$ /java/re/jdk/6_04/promoted/all/b03p/binaries/solaris-sparc/bin/java -showversion -XX:+AggressiveOpts -jar SPECjvm2008.jar -icsv -ict -wt 5s -it 5s -bt 1 compiler.compiler
java version "1.6.0_04-p"
Java(TM) SE Runtime Environment (build 1.6.0_04-p-b03)
Java HotSpot(TM) Server VM (build 11.0-b09, mixed mode)
09-04-2009
EVALUATION
The first failing version is 6u10-b09:
$ /java/re/jdk/6_10/promoted/all/b09/binaries/solaris-sparc/bin/java -showversion -XX:+AggressiveOpts -jar SPECjvm2008.jar -icsv -ict -wt 5s -it 5s -bt 1 compiler.compiler
java version "1.6.0_10-ea"
Java(TM) SE Runtime Environment (build 1.6.0_10-ea-b09)
Java HotSpot(TM) Server VM (build 11.0-b09, mixed mode)
b08 works fine:
$ /java/re/jdk/6_10/promoted/all/b08/binarieses/solaris-sparc/bin/java -showversion -XX:+AggressiveOpts -jar SPECjvm2008.jar -icsv -ict -wt 5s -it 5s -bt 1 compiler.compiler
java version "1.6.0_10-ea"
Java(TM) SE Runtime Environment (build 1.6.0_10-ea-b08)
Java HotSpot(TM) Tiered VM (build 1.6.0_10-ea-b08, mixed mode)
09-04-2009
EVALUATION
As Tom stated that it may be the combination of libjvm.so and rt.jar, I ran the benchmark with every libjvm.so between JDK 7 b20 and b53 copied into the failing JDK 6u13 and all runs were valid.
09-04-2009
EVALUATION
I tried every JDK7 build starting from b20 (which uses HS11.0-b06) up to b53 (which uses HS15.0-b04) and there was no problem. Seems this bug is only in HS11.[1-3].
08-04-2009
EVALUATION
I tried the first version imported into the Mercurial repository (12.0-b01) and that one works. So the fix must be somewhere in TeamWare, but I'm not familiar with TeamWare.
07-04-2009
EVALUATION
I could reproduce the failure on dr-good.sfbay, which is a T1, with:
$ /java/re/jdk/1.6.0_10/latest/binaries/solaris-sparc/bin/java -XX:+AggressiveOpts -jar SPECjvm2008.jar -icsv -ict compiler.compiler
That version fails:
$ /java/re/jdk/1.6.0_13/latest/binaries/solaris-sparc/bin/java -version
java version "1.6.0_13"
Java(TM) SE Runtime Environment (build 1.6.0_13-b03)
Java HotSpot(TM) Server VM (build 11.3-b02, mixed mode)
and this one works:
$ /java/re/jdk/1.6.0_14/latest/binaries/solaris-sparc/bin/java -version
java version "1.6.0_14-ea"
Java(TM) SE Runtime Environment (build 1.6.0_14-ea-b04)
Java HotSpot(TM) Server VM (build 14.0-b13, mixed mode)
07-04-2009
EVALUATION
It doesn't look like a javac bug for the following reasons:
*) It's only reproducible on specific hardware - quoting the submitter "We have two netra_t5220 failing, the other machines are working fine. "
*) If the -XX:+AggressiveOpts is removed from the command line the benchmark succeeds without any single compiler crash
*) specjvm2008 comes with its own version of javac.jar which differs consistently from ours (the bundled javac.jar causes 90 compiler regression tests to fail)
*) The failure described in this reports seems also to depend on which particular jdk is used to run the benchmark. Here's the jdk used by the submitter:
oaf559# /usr/jdk/jdk1.6.0_10/bin/java -version
java version "1.6.0_10"
Java(TM) Platform, Standard Edition for Business (build 1.6.0_10-b33)
Java HotSpot(TM) Server VM (build 11.0-b15, mixed mode)
I tried with other JDK (7b13, 7b24, 7b33, 7b37, 1.6.0_07) and I was *not able to reproduce the bug. It seems like something is broken with this specific JDK *and* this particular Hotspot flag *and* this particular architecture.