Re: Using QRPLOBJ Library

[CRPence <CRPbottle@xxxxxxxxx>]

On 06-Jul-2010 06:52, Charles Wilt wrote:
> I could be mistaken, but IIRC there were some problems with some
> of the larger shops using up permanent addresses with lots of
> create / deletes; which required a reload of the LIC/OS(?) to
> recover from.
>
> That of course didn't go over well, and IBM ended up making some
> changes in a v5 release somewhere that as I understand it allow
> for permanent addresses to be cleaned up for reuse at IPL.
>
> Again, I could very well be mistaken...
>
>
> On Mon, Jul 5, 2010 at 2:51 AM, CRPence wrote:
> > > (well the "not reused" is not completely true anymore; but
> > > it used to be and for most purposes you can still consider it
> > > to be true.)
> >
> > <<SNIP>>
> >
> > I am not sure to what "'not reused' is not completely true
> > anymore" alludes.?

  I recall there was a change to effect a massive increase in the 
available permanent addresses.  I do not recall if that was with 
RISC or a change since\on a newer RISC version.

  And while I believe that there was also a means made available to 
reclaim used permanent addresses in order to avoid a scratch install 
of a system to recover, I am very skeptical the feature is part of 
any normal IPL.  I only recall a /service procedure/ to effect that 
on systems in dire need.  Perhaps instead of having whatever IBM 
support action was performed to activate the reclaim of permanent 
addresses, the IPL code runs such a cleanup automatically under some 
specific conditions; perhaps only after the SRC for no permanent 
addresses available.?

  IMO any such reclaim activity would have to be a very *rare* 
event, because there are potential negative and totally 
unpredictable consequences for the OS code.  If by some dark magic 
[i.e. the OS is not aware that the LIC storage management has 
reincarnated some of its stored-but-destroyed pointers] a stored 
pointer may suddenly appear instead to be a valid object, there is 
no predicting what OS code might do.  Of course that the "object" 
only allows the supported methods would limit most improper accesses 
does give some peace-of-mind, but any "space" [object] can be 
written to... or worse.  Because the effects can not be predicted is 
what makes the rarity of any such reclaim imperative; depending on 
the improbability of negative effects would be a poor design assumption.

Regards, Chuck