[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference turris::ada

Title:DEC Ada
Notice:Ada is no longer a trademark of the US Government
Moderator:KMOOSE::CMCCUTCHEON
Created:Mon Jan 27 1986
Last Modified:Fri Jun 06 1997
Last Successful Update:Fri Jun 06 1997
Number of topics:3874
Total number of notes:16668

3865.0. "ADARTL+0002BB08 %SYSTEM-F-ACCVIO" by AXCL01::BRUTTIN (Armand Bruttin MCS Geneva) Wed Apr 23 1997 13:02

    Hi,
    
    Since the system has been updated to OpenVMS Alpha Version 7.1 (A2100)
    A fortran program calling a DEC ADA package gets the following error
    message during image rundown. ( rundown means: after the END statement
    of the fortran program ).
    
    The customer tries to rebuild its ADA Package and he discovers that
    the LKI_TYPE definition in STARLET_.ADC is not correct. The size of the
    record LKI_TYPE is 56 bytes and the system service $GETLKI with
    LKI$_LOCKS returns 24 bytes per structure.
    
    Even after he has modified the LKI_TYPE and has rebuilt the ADA package
    using DEC ADA V3.3-23 he gets the same original problem.
    
    The same program works as expected on a system running OpenVMS V6.2.
    
    A backup saveset is available on AXPMCS::RMBUG.BCK that contains
    the ADA PACKAGE, the FORTRAN program, a README.TXT and command
    procedures to build the PACKAGE.
    
    
    Please Help !!! 
    
*********************************************************************
    AXAONL$ qptn
    
    Partition    Process     ROCs in use
    ALEPH        AXADAQ::P_ALEP_0  L1RP01 MATS01 VDRPEA 
                                   TPRP03 TPRP04 TPRP05 
                                   TPRP10 TPRP11 TPRP12
                                   TPRP17 TPRP18 TPRP19
                                   TPRP24 TPRP25 TPRP26
                                   TPRP31 TPRP32 TPRP33
    ECAL         AXAONL::P_ECAL_0  ECRPB2 ECRPB1 ECRPB3
                                   ECRPEB

    %SYSTEM-F-ACCVIO, access violation, reason mask=00,
    virtual address=FFFFFFFF00EC42C0, PC=0000000000393B08, PS=0000001B
    %TRACE-F-NOMSG, Message number 0009804C
    %SYSTEM-F-ACCVIO, access violation, reason mask=00, virtual
    address=FFFFFFFF00EC42C0, PC=0000000000393B08, PS=0000001B
    %TRACE-F-NOMSG, Message number 0009804C
    image    module    routine    line      rel PC           abs PC
     ADARTL                          0 000000000002BB08 0000000000393B08
     ADARTL                          0 0000000000054710 00000000003BC710
                                     0 FFFFFFFF80046CC0 FFFFFFFF80046CC0
     PTHREAD$RTL                     0 000000000004D260 00000000004C9260
     PTHREAD$RTL                     0 000000000004C910 00000000004C8910
     PTHREAD$RTL                     0 0000000000030664 00000000004AC664
                                     0 FFFFFFFF869250D8 FFFFFFFF869250D8
    -----------------------------------------------------------------------
    
Thanks,
    
Armand Bruttin.	
T.RTitleUserPersonal
Name
DateLines
3865.1Should be able to recompile Startlet; 7.1 not supported yetBEGIN::CMCCUTCHEONCharlie McCutcheonThu Apr 24 1997 10:3928
>    The customer tries to rebuild its ADA Package and he discovers that
>    the LKI_TYPE definition in STARLET_.ADC is not correct. The size of the
>    record LKI_TYPE is 56 bytes and the system service $GETLKI with
>    LKI$_LOCKS returns 24 bytes per structure.
>    
>    Even after he has modified the LKI_TYPE and has rebuilt the ADA package
>    using DEC ADA V3.3-23 he gets the same original problem.

Please note that Ada V3.3-23 is an ECO kit which changes the passing
mechanism for strings, which may effect how Ada and Fortran communicate.
Read the release notes for more information.

DEC Ada does not specifically support OpenVMS V7.1 on Alpha at this time.
We have a small number of issues we're looking at before providing this
support.

We definitely have not looked at updating our Startlet bindings for 7.1
yet.  The problem you report here sounds fixable by recompiling the
Starlet bindings as you have.  Note that you can ACS EXTRACT SOURCE, then
edit and compile STARTLET_.ADA with your changes.

We're working on several other issues at the moment.  Please mail me
directly if this is a hot issue.  However, note that we will not do much
for a OpenVMS Alpha 7.1 specific problem at this time.

Charlie

    
3865.2KMOOSE::CMCCUTCHEONCharlie McCutcheonFri Apr 25 1997 14:283
We've taken this discussion offline...

Charlie
3865.3KMOOSE::CMCCUTCHEONCharlie McCutcheonTue Apr 29 1997 15:3736
Due to the following mail, I assume the problem is seen with the C program,
and this is not an Ada issue.

From:	AXPMCS::BRUTTIN      25-APR-1997 11:53:39.33
To:	BEGIN::CMCCUTCHEON
CC:	
Subj:	re: May be ADARTL is not the culprit

Charlie,

I not sure if the culprit is ADARTL. The same program written IN C works
fine.

Have I nice weekend,
Armand Bruttin

----------------------------------------
#include stdio
extern int BUG_RETURN_1();

int pm()
{
        int status;
        status = BUG_RETURN_1();
        return status;
}

main()
{
        int status;
        printf("Start C Program\n");
        status = pm();
        printf("End C Program : status = %x", status);
}
--------------------------        

3865.4AXPMCS::BRUTTINArmand Bruttin MCS GenevaThu May 01 1997 06:1624
    Charlie,
    
    Right. The C program does not show the problem.
    
    The problem seems to be an incompatiblity between FORRTL and
    ADARTL!!!
    
    I receive from FORTRAN group a new non-threaded RTL (DEC$FORRTL.EXE_v61 ).
    Even after installing this new RTL the original customer program
    shows the same accvio error. 
    
    Answer from fortran group
    -------------------------------------------------------------------
    <<< Note 1272.13 by QUARK::LIONEL "Free advice is worth every cent" >>>
    
    Given that we don't have the ADARTL sources, I suggest you copy this
    over to the ADA notesfile and ask for assistance.  If it fails using our
    non-threaded RTL, something else is going wrong.
    
                                Steve
   ----------------------------------------------------------------------
    
    
    Armand.
3865.5re: Any news???AXPMCS::BRUTTINArmand Bruttin MCS GenevaTue May 06 1997 12:577
    Charlie,
    
    Any news about this problem?
    
    Thanks,
    
    Armand.
3865.6Still working at itKMOOSE::CMCCUTCHEONCharlie McCutcheonTue May 06 1997 15:3910
Mike's still looking at it.

We don't have much expertease (left) on creating sharable images using Ada
objects.

Last update I heard was that the Ada runtime is apparently not getting
correctly initialized, which would certainly wreak havoc in an Ada program!
As to why, however...  ;-)

Charlie
3865.7Reported as CLDKMOOSE::CMCCUTCHEONCharlie McCutcheonTue May 13 1997 12:131
This problem has been reported as IPMT ZUO101172.