Build of jdk-7udev: Error occurred during initialization of VM

[Pete Brunet]

I just finished building the full forest and all is well.

On 11/15/12 12:17 AM, David Holmes wrote:
> Whoops! Sorry Kelly - Missed that line involving the java.lang.invoke.
>
> David
>
> On 15/11/2012 3:52 PM, Tim Bell wrote:
>> See below for the 'flag day' message related to this, sent to the
>> hotspot and 7u dev lists.
>>
>> On 11/14/12 21:24, David Holmes wrote:
>>> As general advice yes you need matching jdk and hotspot builds in 7u
>>> now (and 8), but this error in itself seems unrelated to invokeDynamic.
>>
>> [snip...]
>>
>>> Pete Brunet wrote:
>>>> I recloned hg.openjdk.java.net/jdk7u/jdk7u-dev/jdk today and the build
>>>> failed as follows:
>>>>
>>>> make[3]: Entering directory
>>>> `/cygdrive/c/Users/Pete/JDK7u/jdk7u-dev/jdk/make/com/sun/jmx'
>>>> /usr/bin/mkdir -p
>>>> ../../../../build/windows-i586/classes/javax/management/remote/rmi
>>>> rm -f
>>>> ../../../../build/windows-i586/classes/javax/management/remote/rmi/RMIConnectionImpl_Stub.class
>>>>
>>>>
>>>> ../../../../build/windows-i586/bin/java -XX:-PrintVMOptions
>>>> -XX:+UnlockDiagnosticVMOptions -XX:-LogVMOutput -client -Xmx512m
>>>> -Xms512m -XX:PermSize=32m -XX:MaxPermSize=160m -cp
>>>> ../../../../build/windows-i586/classes sun.rmi.rmic.Main -classpath
>>>> "../../../../build/windows-i586/classes" \
>>>> -d ../../../../build/windows-i586/classes \
>>>> -v1.2 \
>>>> -keepgenerated \
>>>> javax.management.remote.rmi.RMIConnectionImpl
>>>> Error occurred during initialization of VM
>>>> java/lang/NoClassDefFoundError: java/lang/invoke/AdapterMethodHandle
>>>> make[3]: ***
>>>> [../../../../build/windows-i586/classes/javax/management/remote/rmi/RMIConnectionImpl_Stub.class]
>>>>
>>>>
>>>> Error 1
>>>> make[3]: Leaving directory
>>>> `/cygdrive/c/Users/Pete/JDK7u/jdk7u-dev/jdk/make/com/sun/jmx'
>>
>> This may not be quite the same error fingerprint as described in the
>> flag day message, but having java/lang/invoke involved makes me believe
>> the problems are similar.
>>
>> The short answer is to build the full forest, at least until a jdk7u
>> build is promoted with the JSR292 (invoke dynamic) changes included.
>> After that, you can try using it as an import JDK.
>>
>> Regards-
>>
>> Tim
>>
>> -------- Original Message --------
>> Subject: flag day
>> Date: Wed, 07 Nov 2012 16:53:05 -0700
>> From: Alejandro E Murillo <alejandro.murillo at oracle.com>
>> Organization: Oracle Corporation
>> To: jdk7u-dev at openjdk.java.net, hotspot-dev at openjdk.java.net
>>
>>
>>
>> Many apologies for the delay sending this info
>>
>> This afternoon I integrated hs24-b24 into jdk7u12 [1].
>> That integration includes the changes for JSR292 (invoke dynamic)
>> that were integrated into jdk8 back in August.
>> These changes require the hotspot and jdk repos to be in sync.
>> (hotspot and jdk repos must match - i.e., either both must have the
>> JSR292 changes or
>> both must not have them)
>>
>> A JDK with mismatched repos may build but will not be able to handle
>> invoke dynamics appropriately, and will fail with a message similar to
>> this:
>>
>> $ java -showversion -Xbootclasspath/p:$PWD/classes
>> -XX:+UnlockDiagnosticVMOptions -XX:+EnableInvokeDynamic
>> Invalid layout of java.lang.invoke.MemberName at vmindex
>>
>> In particular, users of JPRT who have pulled the changes must build
>> *both* hotspot and jdk until the JSR292 changes are in a promoted
>> jdk7u12 build (expected tomorrow, Thu, November 8, 2012).
>>
>> The simplest coping strategy is to avoid pulling
>> from the jdk7u master repo until the next jdk7u12 build is promoted.
>> If you do pull, make sure to pull both hotspot and jdk, and to build
>> both.
>>
>> [1] http://hg.openjdk.java.net/jdk7u/jdk7u
>>
>> Thanks