[Info-vax] Java 8: Looking for jna & 64-bit libffi to make jruby work

[Rod Regier]

Hi,

Is there a libffi and jna that will work with the Java 8 kit from HPE at https://www.hpe.com/global/java/download/ivms/1.8.0/jdk8.0_down.html ?

-	John Malmberg made reference to a work in progress for libffi here https://groups.google.com/d/msg/comp.os.vms/IThxF57t9PU/81XzUtIsAgAJ
 but we haven't been able to find those materials. And even if we did, we'd also need a working jna which we can't find anywhere (all extant links back to the work by Philippe Vouters are now dead).
-	Our understanding is we'll make no headway running non-Java modules (needed by our JRuby code) with Java 8 until we have a Java-compatible libffi_shr64.exe jnidispatch.exe + jna.jar (all of which must be rebuilt from source for 64-bit support) as per the following, but all we can find (e.g. from latest VMS Python kit) is a 32-bit version of libffi that doesn't work with Java 8.
-	We get an ACCVIO trying to use current (works with Java 6) libffi + jnidispatch + jna.jar. See background & test results below.

H. Becker & Cedric P. have confirmed here that 32-bit non-Java extensions won't work with Java 8 which uses 64-bit addressing, unlike Java 6, and the likely result will be an ACCVIO error (confirmed also by Oracle):

https://community.hpe.com/t5/Operating-System-OpenVMS/Java-8-and-JDBC-on-OpenVMS-8-4-IA64/m-p/6975305/highlight/true#M37716

Our experience aligns with this: even the simplest test case for JNA fails with Java 8 with an ACCVIO as per:

V:JNA.EXAMPLES> java -version
java version "1.8.0.03-OpenVMS"
Java(TM) SE Runtime Environment (build 1.8.0.03-vms-rc1)
Java HotSpot(TM) 64-Bit Server VM (build 25.51-b02, mixed mode)
V:JNA.EXAMPLES> java -cp .:../jna.jar "DECC_test"
Platform.is64Bit() = true
Platform.isOpenVMS() = true
%SYSTEM-F-ACCVIO, access violation, reason mask=00, virtual address=0000000000000000, PC=0000000000FA77A1, PS=0000001B

  Improperly handled condition, image exit forced by last chance handler.
    Signal arguments:   Number = 0000000000000005
                        Name   = 000000000000000C
                                 0000000000000000
                                 0000000000000000
                                 0000000000FA77A1
                                 000000000000001B

    Register dump:
    R0  = 0000000000000000  R1  = 000000000391C000  R2  = 0000000000000000
    R3  = 0000000000000020  R4  = 0000000000315400  R5  = 000000007A7D7400
    R6  = 000000007A5DE000  R7  = 000007FDC0FD4000  R8  = 0000000000000015
    R9  = 0000000000000015  R10 = 0000000000000065  R11 = 0000000000000074
    SP  = 000000007A7D65C0  TP  = 000000007B24AEC0  R14 = 0000000003735E40
    R15 = 00000000038C61C0  R16 = FFFFFFFF82B9F130  R17 = 0000000000000063
    R18 = 0000000000000072  R19 = 00000000033CDF2F  R20 = 0000000000000061
    R21 = 0000000000000061  R22 = 0000000000000001  R23 = 000000000000003A
    R24 = 0000000000000020  R25 = 0000000000000003  R26 = 0000000000000063
    R27 = 0000000000000065  R28 = 000000000000003A  R29 = 0000000000000020
    R30 = 0000000000000000  R31 = 0000000000000053  PC  = 0000000000FA77A1
    BSP/STORE = 000007FDBFFD5EF0 / 000007FDBFFD5E08 PSR = 000012130842E030
    IIPA = 0000000000FA77A0
    B0  = 0000000000FA7790  B6  = FFFFFFFF82B9F130  B7  = FFFFFFFF800E9620

    Interrupted Frame RSE Backing Store, Size = 16 registers

    R32 = 000000007A7D6F80  R33 = 0000000000000000  R34 = C000000000000389
    R35 = FFFFFFFF803676F0  R36 = 000000000391C000  R37 = 0000000004AA37B0
    R38 = 000000007A7D6710  R39 = 000000000000000C  R40 = 00000000033CDB20
    R41 = 0000000000000000  R42 = 0000000000000000  R43 = 0000000000000020
    R44 = 0000000000000001  R45 = 00000000033CDB20  R46 = 0000000000000020
    R47 = 0000000000000015

Thanks,

Rod Regier & Ben Armstrong, Dymaxion Research Ltd.