[top posting]

http://www.trainersfriend.com/Language_Environment_courses/m512descr.htm

is multi-lingual: COBOL, PL/I, C, Assembler - take your choice


[You forgot to set your Reply-to back to the list]



On 6/12/2012 3:20 PM, Frank Swarbrick wrote:
So it sounds like the original author should have never used R15 in the first 
place for the base register.  OK.

The program doesn't use any macros other than the CEE ones I just added and 
ABEND.  But I'll see if I can find a free register other than 14 (or 15 or 0 or 
1!).  It's pretty much used every other register at some point in the program, 
but hopefully I can get around that.

I've been reading about "baseless" programming and all new code I've added uses 
it, but at this point it don't want to make too many changes to it.  So is your 
suggestion to use RC=RETCODE,MODIFIER=MOD and have those two as part of my automatic 
storage area?  Actually, if I do that I can probably get away (for now) with leaving it 
using R15, since those fields will be based on R13.

I will try this out.  Thanks for your help!!!!
Frank




________________________________
From: "McKown, John"<[email protected]>
To: [email protected]
Sent: Tuesday, June 12, 2012 2:53 PM
Subject: Re: CEETERM and register 15

The reason for the L in CEETERM is because you can do CEETERM 
RC=VARNAME,MODIFIER=MODNAME where VARNAME and MODNAME are fullword areas.

In general, in z/OS, you should __never__ use R0 (impossible actually), R1, 
R14, or R15 for a CSECT/RSECT base register. Many (most) IBM macros use these 
registers. And occassionally other registers as well.

For LE enabled assembler (I have written some), when I use a base register at all (not often any 
more), I use R11 for my first, then R10, R9, and so as as the module grows in size. In 
"modern" programming, you can avoid the use of a "base register" for code in 
almost every case by using relative instructions instead of based instructions. E.g. use LARL 
instead of LA for areas in the CSECT, and Jxx instead of Bxx for branching. If you load constant 
values, try to use LHI if possible. Do try to write reentrant code. If you do that, you can keep 
the modifiable values in the DSA, which is addressed by R13 and which is set up for you with the 
CEEENTRY and CEEDSA macros. For a constant which does not fit into a halfword, or maybe is packed 
decimal or characters, you can replace somethin like: L R7,FULLWORD with 2 instructions: LARL 
R7,FULLWORD followed by L R7,0(,R7), and the same with packed and characters. OK, that is a bit 
ugly, I admit.


--
John McKown
Systems Engineer IV
IT

Administrative Services Group

HealthMarkets®

9151 Boulevard 26 . N. Richland Hills . TX 76010
(817) 255-3225 phone .
[email protected] . www.HealthMarkets.com

Confidentiality Notice: This e-mail message may contain confidential or 
proprietary information. If you are not the intended recipient, please contact 
the sender by reply e-mail and destroy all copies of the original message. 
HealthMarkets® is the brand name for products underwritten and issued by the 
insurance subsidiaries of HealthMarkets, Inc. -The Chesapeake Life Insurance 
Company®, Mid-West National Life Insurance Company of TennesseeSM and The MEGA 
Life and Health Insurance Company.SM

-----Original Message-----
From: IBM Mainframe Assembler List
[mailto:[email protected]] On Behalf Of Frank Swarbrick
Sent: Tuesday, June 12, 2012 3:25 PM
To: [email protected]
Subject: CEETERM and register 15

First off please realize that I'm an assembler beginner.
I've fiddled with it here and there for 15 years, but I'm
really just a COBOL programmer.  So please be gentle!  :-)

We have a vendor written, code supplied, subroutine that does
for us a few things that COBOL can't (bit fiddling; posting
of an ECB; etc.).  This same exact routine is called both by
batch COBOL program and CICS COBOL programs.  It is not
reentrant.  In order to make it reentrant I have been working
on "LE-enabling" it.  This seems to work quite well!  But I
ran in to something "weird" with CEETERM.  CEETERM RC=0
generates the following:

58F0 F098            013A0  1827+         L   15,=A(0)

5800 F098            013A0  1828+         L   0,=A(0)
58DD 0004            00004  1829+         L   13,4(13)
58E0 D00C            0000C  1830+         L   14,12(,13)
981C D018            00018  1831+         LM  1,12,24(13)
07FE                        1832+         BR  14

Problem?  The program uses register 15 as the base register.
Funny thing is that the code actually "works".  But it really
doesn't. The first L sets R15 to 0.  Then the second L loads
in to R0 from 152(,R15), but of course R15 is now 0, not the
real base address it needs to be.  It's only coincidence that
152 bytes in to low core contains x'00000000'.  If I tried to
use an RC of some other value it would still be set to 0.


I changed the program to use 14 instead of 15 for the base
register and all is now fine.  But I don't see it documented
that CEETERM should not be used when the base is R15.  Or is
it just common practice to not use R15 as the base?  Other
than this one program I don't see that R15 is generally used
for that purpose even though it starts out with the correct
address, which is probably why it was used in this case.
Specifically, the original code was this:
SUBR     CSECT

           USING *,RF
           B     START
           DC    CL8'SUBR'
START    DS    0H
           SAVE  (14,12)

Anyway, I have to wonder why CEETERM doesn't take a page from
the RETURN macro, and use LA from an offset into "R0", i.e.,
if I had CEETERM RC=12,MODIFIER=8, we'd get something like this:

LA  15,12(0)
LA  0,8(0)
L   13,4(13)
L   14,12(,13)
LM  1,12,24(13)
BR  14

No base register needed, correct?  (This is one of those
cases where "register 0" is really just "zero", right, no
matter what value it actually holds?)


(Apologies to Russ Evans, if he is reading this, for taking
an unexpected side trip to LE land.)

Hope I've not shown too much ignorance.


Thanks!
Frank








http://www.trainersfriend.com/Language_Environment_courses/m512descr.htm

is multi-lingual: COBOL, PL/I, C, Assembler - take your choice


--

Kind regards,

-Steve Comstock
The Trainer's Friend, Inc.

303-355-2752
http://www.trainersfriend.com

* To get a good Return on your Investment, first make an investment!
  + Training your people is an excellent investment

* Try our tool for calculating your Return On Investment
    for training dollars at
  http://www.trainersfriend.com/ROI/roi.html

Reply via email to