Path: utzoo!utgpu!jarvis.csri.toronto.edu!rutgers!ucsd!tut.cis.ohio-state.edu!purdue!haven!udel!udccvax1!mikej
From: mikej@vax1.acs.udel.EDU (Michael Jacobs)
Newsgroups: comp.sys.ibm.pc
Subject: Re: chaining interrupts
Keywords: chaining interrupts
Message-ID: <5026@udccvax1.acs.udel.EDU>
Date: 16 Nov 89 04:58:28 GMT
References: <4543@solo5.cs.vu.nl>
Reply-To: mikej@vax1.acs.udel.EDU (Michael Jacobs)
Organization: Yoyodyne Propulsion Systems
Lines: 64

>	I have written several own interrupt handlers for the ibm pc,
>and I wonder if it's possible to use the old interrupt routines by
>chaining from the new one. One way to do this is by setting the old
>vector on the address of an other interrupt number. A disadvantage
>of this method is that you don't always know which interrupt is free
>and you can run out of all possible interrupts.
>	So is there an other way, from TP4 for instance:
>		jmp oldvector (with inline code) 
>doesn't succeed.

I was playing with interrupts last night and just worked it out.  I have
succeeded in replacing interrupt $1C (the clock updater), and then, inside
my routine, calling the clock updated (so it gets done), then returning
safely.

in turbo 5.5 (it would work in tp >= 4.0 I would think):

{$m 1024,0,0}    { this program installs the interrupt replacement,
				   and since we want the code to stay in memory
				   so it can be used, we'll have to use the KEEP
				   command, so we don't want a big stack and heap
				   clogging memory }

uses dos;   { getintvec and setintvec are in unit dos }

var ticksave : pointer;    { this holds onto the old interrupt vector }

	procedure tick; interrupt;  { this is what would get called instead }
	begin
	inline ( $50/$bb/ticksave/$ff/$1f )  { call tick routine }
                                  { what this does is
								   PUSH AX
								   MOV BX, OFFSET ticksave
								   CALL [BX] (intersegment call) 

								   I push AX, because at the end of
								   the old clock updating stuff, there
								   will be an IRET which pops the
								   offset (IP), the segment (CS), and
								   the flags.  since the flags are 
								   irrelevent in this case, i just
								   pushed the garbage in AX }
	 
   { your code here... }
	end;

	var X : pointer;
	begin
	getintvec ( $1C, ticksave );   { get the old vector and remember it }
	x := @tick;                    { the address of tick is the new vector }
	setintvec ( $1c, x );          { set it }
	keep ( 0 )                     { end the program but keep all this }
	end.                           {   BS in memory }

	 
fun, huh?



-- 

Mike J             |  
The Grey Sysop...  |  Phone...RING!...yep yep yep yep yep!
                   |