From: Andy.Tainter@f615.n109.z1.fidonet.org (Andy Tainter) Subject: BUG in 0.99[p1] kernel c Date: Mon, 04 Jan 1993 00:00:00 -0800
JS>From: jes@grendel.demon.co.uk (Jim Segrave)
JS>In article <725836861.AA28902@remote.halcyon.com>
JS>Andy.Tainter@f615.n109.z1.fidonet.org (Andy Tainter) writes:
JS>>You *CAN* have more than one card on INT 2, this is a cascaded int
and
JS>>you can have upto 7 (may be 6) devices using it...
JS>>
JS>>That is DOS, maybe linux cannot do that, but the hardware is set
upto
JS>>do
JS>>it...
JS>This is not quite correct. INT 2 is connected to the interrupt
request
JS>output of the second interrupt controller which in turn is connnected
to
JS>the ISA bus IRQ8..IRQ15 lines. When one of these lines has an active
JS>high edge, the second controller generates an interrupt request
which
JS>then activates the INT 2 input of the master interrupt controller.
This
JS>input is NOT connected to any line anywhere on the ISA bus. The
JS>interrupt controllers are programmed for cascade operation on the
master's
JS>INT 2 request line so that the interrupt vector passed to th CPU on
an
JS>interrupt generated on IRQ8..IRQ15 will get its vector supplied from
the
JS>second interrupt controller. Of course Linux upports this, if it
didn't,
JS>no non-SCSI disc would work.
JS>None of this has anything to do with sharing a bus IRQ line between
cards -
JS>the advice posted earlier is correct - you can not share interrupts
JS>between cards without modifying the motherboard with diode networks.
It
JS>may work at first but the long term effect is likely to be component
damage.
Interesting, we have done it on a couple of really loaded PC's at work
with no apparent problems...
Andy
___
X SLMR 2.1a X This tagline stolen by Silly Little Mail Reader!
* Origin: Enlightened Board (703) 370-9528 (1:109/615)