Lindy,
Implementing a PC routine requires a little bit more planning and consideration
than the SVC method.
For starters, you are going to need a resource-owning address space to own the
linkage index.
You are going to have to understand the implications of providing the service
you intend and consider carefully the normal (and abnormal) removal of the same.
Is the service going to be available to all address spaces or just a select few?
I suggest reading up on any of the following you are not familiar with :
MVCDK and MVCSK
Resource managers
ASN and LX reuse
System integrity
Most of the installation-written SVCs I have come across have been more of a
"smash and grab" solution.
The SVCs were "easier" to write as there was less baggage to accompany them and
they could be expressed in a single program.
Some of the installation-written SVCs I have come across have had system
integrity holes so big you could drive a truck through.
Personally, I would not even consider writing an SVC these days.
Strangely, some sites still have objections to any new address spaces appearing
on the system, even if all it does is sit there and quietly own stuff.
Rob Scott
Lead Developer
Rocket Software
77 Fourth Avenue . Suite 100 . Waltham . MA 02451-1468 . USA
Tel: +1.781.684.2305
Email: [email protected]
Web: www.rocketsoftware.com
-----Original Message-----
From: IBM Mainframe Assembler List [mailto:[email protected]] On
Behalf Of John Gilmore
Sent: 08 February 2013 03:04
To: [email protected]
Subject: Re: SVC 34
What I wrote was:
| The question
|| Were once SVC's ok, and now they are not preferred?
| could be colored naif, but it is an entirely legitimate one, and the
answer is yes.
It reflected my view that, while Lindy's question had been condescended to, it
was in fact an "entirely legitimate" one.
This view, expressed umambiguously, was somehow transmogrified into a vehicle
of insult followed by condescension. It was neither, and a neutral native
English speaker would not have judged it to be either.
Let me therefore make my advice more specific. SVC routines are lumbered with
a long history of machinery---types, transient areas, segmentation into
4096-byte pieces, etc., etc,---that is now irrelevant. If you have SVC
routines that do something that is at once useful and secure, maintain them;
if instead you are setting out to do something ab initio, use PC routines
instead. Finally, try to consider, dispassionately, the uncomfortable question
whether you know enough to do what you are proposing to do safely and securely.
I entirely agree with the substance of what Peter Relson has been saying (if
not always with the wording of his strictures). Services should be implemented
only by people who know what they are doing, i.e., by the over-competent. Most
of the user SVC routines I have examined are 1) dispensable, 2) ill-conceived,
and 3) insecurely implemented.
John Gilmore, Ashland, MA 01721 - USA
