Path: utzoo!utgpu!jarvis.csri.toronto.edu!rutgers!bellcore!texbell!merch!cpe!hal6000!trsvax!don
From: don@trsvax.UUCP
Newsgroups: comp.sys.ibm.pc
Subject: Re: Help from TSR Gurus requested
Message-ID: <216100111@trsvax>
Date: 20 Jul 89 13:55:00 GMT
References: <536@cybaswan.UUCP>
Lines: 32
Nf-ID: #R:cybaswan.UUCP:536:trsvax:216100111:000:1548
Nf-From: trsvax.UUCP!don    Jul 20 08:55:00 1989


>>Now the Question: Can the resident part of a TSR get away with using the area
>>below 5Ch? I would *think* that DOS has no further use for the PSP after the
>>program becomes resident, but of course the area may still figure
>>in some sort of linked-list internally in DOS.

>Once you hooked up your TSR routine to any interrupt handler, the PSP is not
>used anymore.

This is not in general correct.  It may be correct for some TSR's, but if a
TSR performs any interrupt 21h functions (after checking for re-entrancy, of
course), some values in the current PSP are altered, so you need to set up
your PSP as the current PSP before performing such operations.  The same
goes for disk I/O.  At any rate, a TSR's PSP may be used after the program
has gone resident, so you need to make sure it isn't used before you re-use
it.

BTW, sometimes I move the stack pointer to 100h in a com file so the stack
grows over the PSP.  If I need more stack, I can make my first statement as
a jump, then allocate a data buffer immediately after that point, setting up
the stack to point at the end of my new buffer.  I assume by the time the 
stack has grown over my first statement, I've already executed it.

---------------------------------------------------------------------
I purchase Product 19 because it's the only cereal I know named after a
prime number.
			- paraphrased from a posting in rec.humor.funny

Don Subt			The opinions expressed above are
Tandy Corp.			strictly mine, not my employer's.

817-390-3068			...!killer!ninja!sys1!trsvax!don