RE: Re: GCIC Fullscreen applications

Mike
 
GENQ  kuku
GDEQ  kuku
 
Which are actually SVC calls to my kernel
 
will enque and deque the  logical resource KUKU
 
Will put any other task that does an enq on that resource into a wait state
until the deque is issued by the enquing task
 
all other tasks will continue to work
 
 
Roc


  _____  

From: [email protected] [mailto:[email protected]] On Behalf Of
mfnoel
Sent: Sunday, August 08, 2010 18:31 
To: [email protected]
Subject: [H390-VM] Re: GCIC Fullscreen applications


  

KICKS is written almost entirely in GCCMVS, and many of it's routines are
used recursively (meaning they call themselves, either directly or in some
chain of calls). And it uses PCPCLIB all over the place (carefully!)
including in the recursively called routines.

One of the reasons it's written almost entirely in GCCMVS is that the
compiler provides pretty good inline assembler capability, meaning you don't
have to resort to another standalone assembler routine everytime you turn
around. Here's quick example, doing an enqueue within a C routine...

if (csa->enqscope[6] == 'S') {
__asm__ (
"L 2,%1\n\t"
"L 3,%2\n\t"
"ENQ ((2),(3),E,44,SYSTEMS),RET=USE\n\t"
"ST 15,%0"
: "=m"(rtn)
: "m"(qname),"m"(rname)
: "0","1","2","3","15"
);
}

I think with some understanding of reuseable/reentrant/recursive coding you
can use GCCMVS almost anywhere you could use assembler. I'd be amazed if the
CMS version broke such functionality -- but of course I'm amazed by stuff
everyday...

--- In [email protected] <mailto:H390-VM%40yahoogroups.com> , "Joe
Monk" <joe.monk@...> wrote:
>
> Paul -
> 
> How is the sort function supposed to work in a multi tasking envrionment,
if
> it cant be *arbitrarily* reentrant?
> 
> Joe
> -----Original Message-----
> From: [email protected] <mailto:H390-VM%40yahoogroups.com>
[mailto:[email protected] <mailto:H390-VM%40yahoogroups.com> ] On
Behalf Of
> kerravon86
> Sent: Sunday, August 08, 2010 7:21 AM
> To: [email protected] <mailto:H390-VM%40yahoogroups.com> 
> Subject: [H390-VM] Re: GCIC Fullscreen applications
> 
> --- In [email protected] <mailto:H390-VM%40yahoogroups.com> , "Joe
Monk" <joe.monk@> wrote:
> >
> > Just give up. He's already admitted that its not doable.
> 
> I admitted that *what* is not doable?
> 
> > >> There would be a lot of extra baggage required to force an 
> > >> arbitrary C program to be reentrant, baggage that is not yet 
> > >> available.
> 
> Forcing an arbitrary C program to be reentrant?
> 
> Yes, quibbling aside, that isn't possible.
> 
> But in the environment he is thinking of using it in, he doesn't need it
to
> be an arbitrary C program. He can write his C code so that it's naturally
> reentrant, and doesn't use library functions, except perhaps the ones that
> are able to be made available in a non-hosted environment.
> 
> BFN. Paul.
> 
> 
> 
> 
> ------------------------------------
> 
> Yahoo! Groups Links
>