Conference clt::dec_pascal_bugs

827.0. "INCORRECT CODE PRODUCED BY DEC PASCAL FOR OPENVMS" by TAV02::MILCHIN () Tue Feb 11 1997 05:28

	HI!

	Bellow is a minimized example provided me by the customer, which
illustrates an incorrect code produced by DEC PASCAL for OPENVMS AXP.
The program fails with the %SYSTEM-F-ACCVIO,... refering to any element
of 2-dimentional array passed as conformant array to a procedure.
Provided is a source code and the DEBUGGER's session log file.

	Environment:
	OPENVMS ALPHA V7.1, DEC PASCAL V5.5-57-329E.

		MICHAEL
******************************************************************
{[align(vax)]}

PROGRAM TEST2dim(input,output);

CONST
    DIM1    =   11;
    DIM2    =   66000;
VAR

    tblr : array[1..DIM1,1..DIM2] of real;

    r   :   real;
    i1, i2: integer;



[GLOBAL]function INTER2DIMR( {no}var tbl : array[la1..ua1:integer;
						la2..ua2:integer]
				     of real): real;
VAR
    LR    : real;

BEGIN
    
    LR := tbl[1,1];

    writeln('  real is: ', LR);
    INTER2DIMR := LR;
END;



PROCEDURE CREATE_2DIMR({no}var tbl :  array[lb1..ub1:integer;
					    lb2..ub2:integer] of real);
BEGIN
    {<<< CREATE TABLE >>>}
{    FOR i1:=1 TO dim1 DO
    FOR i2:=1 TO dim2 DO
	tbl[i1,i2]:= 1.0;
}
	tbl[1,2]:= 1.0;
END;

PROCEDURE P;
BEGIN
    R  :=     INTER2DIMR( tblR);
END;

BEGIN
    CREATE_2dimr(tblR);
{    P;}
END.

	The DEBUGGER session log file:

Step
!stepped to TEST2DIM\%LINE 50
!    50:     CREATE_2dimr(tblR);
step/into
!stepped to TEST2DIM\CREATE_2DIMR\%LINE 41
!    41:         tbl[1,2]:= 1.0;
sh symb/type tbl
!data TEST2DIM\CREATE_2DIMR\TBL
!    noncontiguous array descriptor type, 2 dimensions, bounds: [1:11,1:66000], size: 2904000 bytes
!        cell type: atomic type, F_floating, size: 4 bytes
!data TEST2DIM\INTER2DIMR\TBL
!     (type/address information is not available)
eval/addr tbl
!196616
eval/addr/hex tbl
!0000000000030008
eval/addr/hex tbl[1,2]
!000000000003000C
set break/exception
go
!%SYSTEM-F-ACCVIO, access violation, reason mask=00, virtual address=000000010003000C, PC=00000000000201C4, PS=0000001B
!break on exception at TEST2DIM\CREATE_2DIMR\%LINE 41+92
!    41:         tbl[1,2]:= 1.0;
exam/instr @pc
!TEST2DIM\CREATE_2DIMR\%LINE 41+92:              STF             F0,(R1)
exam/hex r1
!TEST2DIM\CREATE_2DIMR\%R1:      000000010003000C
EXIT

I reproduced this...I'll look a little more closely.

Meg

	HI!

	I just would like to know if there is any progress in testing the
problem reported in .0. 
	REGARDS!
		MICHAEL

I've only looked at it briefly and don't expect to have time
to get back to it for a couple of weeks.  It appears to have
something to do with the very large array dimensions - but that's
just a guess.

Meg

    Just for an update.  I've been working on this for a couple of days
    now.  I'm testing a fix and hope to have it fixed in another day
    or two.  I hope to have a network kit out in a a couple of weeks.
    How critical is this to the customer?
    
    				-John

    OK, after a couple of false starts I got it fixed (I hope!)  The
    fix will be included in the next network kit.
    
    				-John