Path: utzoo!utgpu!news-server.csri.toronto.edu!mailrus!cs.utexas.edu!usc!zaphod.mps.ohio-state.edu!sdd.hp.com!elroy.jpl.nasa.gov!peregrine!ccicpg!legs!kaz
From: kaz@legs.UUCP (Richard Kaczmarek)
Newsgroups: comp.sys.ibm.pc
Subject: Re: What's IRQ 2 used for on an AT?
Summary: IRQ 2 -> IRQ 9
Message-ID: <2865@legs.UUCP>
Date: 31 May 90 17:20:08 GMT
References: <749@ucsvc.ucs.unimelb.edu.au>
Organization: AST Research, ASIC (VLSI) Development, Irvine, CA
Lines: 28

There are two 8259 interrupt controllers in the AT.  I believe the
master is located at address 20h and the slave is at A0h.  The master
handles IRQs 0 through 7 and the slave handles IRQs 8 through 15.  The
master "yanks" the slave via IRQ 2.

From what I understand (I'm a software weenie, not hardware), IRQ 2 is
remapped in hardware to IRQ 9.  Any hardware requests to IRQ 2 yank 
IRQ 9's line.

The processor, by virtue of the IRQ number, uses the interrupt vector 
table to call some code relating to that hardware request.  Although 
IRQ 2's vector should be used, recall it was remapped to IRQ 9; 
IRQ 9's software interrupt vector, then, is jumped to when either an IRQ 9
or a real IRQ 2 occurs.

There's a piece of BIOS code sitting where IRQ 9's interrupt vector 
is pointing which determines who _really_ yanked the line (2 or 9).  If
9, it continues on; if 2, it jumps over to 2's interrupt vector.

That's why your stuff worked.  Although you set it to 2, the motherboard's
hardware remapped it to 9.  The software there took care of it, no
problem.

-- 
Rich Kaczmarek - 			uucp: ...!uunet!ccicpg!legs!kaz
AST Research Inc.			Fax: (714) 727-9358
P. O. Box 19658				Tel: (714) 727-8621
Irvine, California 92713-9658