topic Re: MACRO 32 question in Operating System - OpenVMS

MACRO 32 question

marchunan — Wed, 02 Nov 2011 04:56:05 GMT

Hi all,

This is Archunan. Happy contacting you all as a vsitor again after a long time. I need to learn this renamed forum ot use my previous userID.

I have a question on my MACRO 32 which currently runs on Itanium//OpenVMS 821. The purpose of this simple program is retreves few login sysgem paramter values for reporting purpose. All login parameter ITEM names are declared below in SYSGEN_ITMLST. The function KAN_CPX_SYSGEN moves all the itemlist values ( GETSYIW routine has been called above ) to R6. Please see the descriptor details.

Few related code snagit

ACCOUNT_STRING: .BLKB UAF$S_USERNAME ; Generic account name

.LONG 0

ACCOUNT_INFO: .BLKB 1024 ; Generic account information

.LONG 0

ACCOUNT_INFO1: .LONG 8 ; Generic account information

GETSYI:

$GETSYIW EFN=WEFN,ITMLST=SYSGEN_ITMLST1

SYSGEN_ITMLST:

ITMLST 4,SYI$_LGI_BRK_LIM,LGI_BRK_LIM,0.

....

.LONG 0 ;END OF LIST

SYSGEN_ITEMS_TABLE:

.ADDRESS LGI_BRK_LIM

.ADDRESS LGI_BRK_LIM_TEXT

.ADDRESS LGI_BRK_LIM_DSC

......

SYSGEN_ITEMS_TABLE_SIZE = <.-SYSGEN_ITEMS_TABLE>/12

LGI_BRK_LIM_TEXT: .ASCID /LGI_BRK_LIM/

LGI_BRK_LIM_DSC: .ASCID ! (Brklim) / Login Breakin count!

...

ASCII_NUM_DSC: .WORD 12

.BYTE DSC$K_DTYPE_T

.BYTE DSC$K_CLASS_S

.ADDRESS ASCII_NUMBUF

ASCII_NUMBUF: .BLKB 12

:The function which retrieves the breakin limit....

.ENTRY KAN_CPX_SYSGEN,^M<R2,R3,R4,R5,R6,R7,R8,R9>

MOVAL SYSGEN_ITEMS_TABLE,R6 ;GET TABLE OF ITEMS TO REPORT

MOVL #SYSGEN_ITEMS_TABLE_SIZE,R7

20$:

PUSHAQ ASCII_NUM_DSC ;RESULTANT STRING DESCRIPTOR

PUSHL (R6)+ ;GET VALUE

CALLS #2,G^OTS$CVT_L_TI ;CONVERT LOGIN FAILURES

BLBC R0,30$

MOVL (R6)+,R2 ;GET ITEM DESCRIPTOR

MOVZWL (R2),R0

MOVL 4(R2),R1

MOVC3 R0,(R1),ACCOUNT_STRING ;COPY ITEM NAME

CLRB (R3) ;NULL TERMINATE

MOVZWL ASCII_NUM_DSC,R0

MOVL ASCII_NUM_DSC+4,R1

MOVC3 R0,(R1),ACCOUNT_INFO ;COPY VALUE

MOVL (R6)+,R2 ;GET DESCRIPTION

MOVZWL (R2),R0

MOVL 4(R2),R1

MOVC3 R0,(R1),(R3) ;APPEND DESCRIPTION

CLRB (R3) ;NULL TERMINATE

............................................................... <New code> to be added (given below) to test the value in ACCOUNT_INFO.

CALLS #3,RPT_LOGIN_BRK ;routine to report the ITM lists values

SOBGTR R7,20$ ;DO NEXT ONE...

MOVZWL #SS$_NORMAL,R0

RET

30$:

MOVL R0,R2 ;SAVE ERROR

PUSHL #0

PUSHAB OTSCVT_MSG ;SEND MESSAGE

PUSHL R0 ;SEND ERROR CODE

PUSHL #0

CALLS #4,G^SYSERR_RPT ;REPORT ERROR

MOVL R2,R0 ;RESTORE ERROR

RET

<FF>

My question is that I would like to test & report only when LGI_BRK_LIM value is > 5. I tested this with the following code...

<New code>

MOVAL ASCII_NUM_DSC,R0

MOVZBL (R0),R1 ; len of buf string, converted into dec

MOVL 4(R0),R0 ; get buf address

MOVC3 R1,(R0),ACCOUNT_INFO1 ; move " 6" to the descriptor ACCOUNT_INFO1

.........? Here I need to COMP (or SUB) integer 5 from with the value in ACCOUNT_INFO1. I tried OTS$ routine to convert, " 6" to integer, but still CMP returns always greater than 5. Seems there are some junk characters in the descriptor

BGTR 26$

RET

26$:

call to Report program

Ret

Thanks

Archie

Re: MACRO 32 question

Steven Schweda — Wed, 02 Nov 2011 03:32:29 GMT

> [...] MACRO 32 [...]

   WHY? (If doing things the hard way causes problems, then why not try
the easy way?)

   I don't know enough Macro 32 to do anything, but I can't see your
call to $GETSYI.

   If you want the integer value returned by $GETSYI, then why are you
converting it into a string using OTS$CVT_L_TI ?

> I tried some unsuccesful ways to extract this string 5 as a number for
> comparison.

   Wasn't it a number before you started playing with it?

Re: MACRO 32 question

abrsvc — Wed, 02 Nov 2011 12:11:01 GMT

The code shown is oncomplete, but I'd agree with Steven. You already have the value prior to the conversion to a sring. Perhaps the string form is needed for the output, but save the initial value in a register that is not used and test for thatt later. For example, something like this:

Movzbl (r6),r10

remaining code to convert to strings (account info)

cmpb R10,#5

BLEQ 10$

calls #3, reporting routine

10$:

Re: MACRO 32 question

Hoff — Wed, 02 Nov 2011 12:18:49 GMT

Please post your complete example. It appears the do-not-care event flag isn't being used here, for instance, and there does not appear to be an IOSB specified and verified. If you're seeking to learn Macro32, then there are various other coding errors here, in other words.

Is there a reason why a language as primitive and comparatively expensive as assembler is in use here? (It'd be more effective rewrite this sequence in C than to debug this case, this even given C isn't all that much better than assembler in terms of its maintenance and features. (This usually implies you're not doing this assembler work commercially, or that there's a large base of Macro32 here and no plans to migrate.)

If you're working in a business environment that lacks access to compilers, then python, Perl, DCL or one of the other "free" languages would be a more effective use of your time. (I would hesitate to suggest Bliss, but that's also free.) The DCL lexical functions including f$getsyi can fetch these details, for instance.

There's C code to fetch various information from SYSUAF using $getsyiw calls in this C source code library, too.

There are other source code examples around.

And OpenVMS I64 V8.2-1? That's a very old and rather sketchy release; is there a reason why you're using that release and not something more current? There are latent errors in the older versions of the Macro32 compiler on OpenVMS I64.

Re: MACRO 32 question

John Gillings — Wed, 02 Nov 2011 21:08:07 GMT

Archunan,

I'm a bit confused by your problem description. You say you want to test LGI_BRK_LIM? It appears you have a variable called LGI_BRK_LIM referenced in the item list for $GETSYI (there is no definition for the macro ITMLST, so I can't be certain).

If that's the case, the code should be

CMPL LGI_BRK_LIM,#5

BLEQ SkipReport

... code for report

Most of the complexity here seems to involve conversions, formatting and building output strings. I find the easiest way to deal with this in MACRO32 is using $FAO.

Re: MACRO 32 question

marchunan — Tue, 15 Nov 2011 14:24:33 GMT

Steven, I have called $GETSYI before come to this procedure and got the required sysgen values. And as you suggesed I tested the values before converting also. Thanks Steven.

Re: MACRO 32 question

marchunan — Tue, 15 Nov 2011 14:27:09 GMT

It should work as per the suggested code, but I see CMP didn't work as R10 didn't have correct value. Thanks abrsvc

Re: MACRO 32 question

marchunan — Tue, 15 Nov 2011 15:09:41 GMT

Hi Hoff, thanks for your response. This is very huge applciation. the initial developer is one of the greatest VMS programmer on my view. I wonder seeing his code. He must be from DEC engg team or DEC engg team have helped him from the shoulder.We don't have any idea of converting to C. V8.2-1 is my build server and I use this version as per HP eng. suggestion.

Re: MACRO 32 question

marchunan — Tue, 15 Nov 2011 15:36:46 GMT

Sorry for the delayed response. somehow the issue been resolved and I completed the build also.

Thanks John, Steven, Hoff, & abrsvc.

I didn't past the itmlst declaration, desc, getsyi call,,,. tested SYI$_LGI_BRK_LIM, but somehow I miss moving the getsyi return values properily between stack and register on the calls to $GETSYI, LIB$TRAVERSE_TREE, and $FAO. And I had difficulties looking into the register values during the debug. Anyway the issue been resolved with the help of quick small c program call.

Thanks to everyone for your valuable time.

Thanks
Archunan

Re: MACRO 32 question

marchunan — Tue, 15 Nov 2011 15:44:48 GMT

This issue has been resolved, I don't know how to close this topic. I will close shortly.

Re: MACRO 32 question

Hoff — Tue, 15 Nov 2011 20:10:44 GMT

Even OpenVMS Engineering moved off of using Macro32 for new development probably a dozen years ago, or more. I replaced more than a little Macro32 code back when I was working in engineering, where somewhat more than spot maintenance or tweaks was required.

If HP suggested OpenVMS I64 V8.2-1, then there is other information not in evidence or some sort of a kernel or API dependency that we don't know about, or this version recommendation is stale information. V8.2-1 is old, comparatively unstable and not something I would expect HP (or anybody else) to recommend.

Macro32 written by the even best engineers is still ancient, complex, hard to extend, slow to program and comparatively expensive to support. This is not 1985, after all. The assembler langiuage wars are long over, and the compilers (and now the interpreters and JIT) have won.

If anything, C is far too low level by present standards, but VMS doesn't have particularly robust support for newer languages and tools.

Re: MACRO 32 question

John McL — Wed, 16 Nov 2011 22:22:26 GMT

marchunan, I believe that to close a thread you have to login then find the "this is the solution" option associated with a specific posting by someone.

As far as I could discover, there's no way to close a post other than to say that some posting provided the solution.

This seems to be yet another failing of this system. Sometimes a matter is raised for discussion or to highlight an issue, as I did recently with a documentation flaw. A solution is not always required, might not be given and while waiting for someone else to point you in the right direction you might have solved it yourself.