[plug] eth0 and ppp

[Garth Atkinson]

russ wrote:
> 
> Len Bird wrote:
> > The interrupt trap is usually overlooked by the gurus! I am really a novice
> > in Linux but suggest that you boot up in humble old dos for which there are
> > a number of programs like FINDIRQ, CHECKIT and others which will identify
> > which hardware item is using which Interrupt, and which are unused. INT 9 is
> > paired with INT 2 which therefore needs watching. Most Ethernet Cards these
> > days are software configured, but you may quite well have a Jumper-link IRQ
> > selection. It is usually easier to change the IRQ for the Ethernet Card than
> > for the Com Port. If you are stuck for one of these, I can help.
> >
> > If you are now sure that IRQ conflict  is the problem, perhaps some of the
> > more knowledgeable of the PLUG subscribers can better advise you,
> 
> I'm pretty sure that's the problem. The pcmcia finds both cards and
> assigns IRQ3 to the ethernet card and 9 to the serial card. There are
> no jumpers on the pcmcia cards. I tried forcing the interrupt for the
> card to 9 using ifconfig but it didn't work.
> 
> Thanks
> russ
Hello Russ

If you have the kernel source code installed, edit serial.c to change
the IRQ used for your serial port, and recompile the kernel.

I not sure what Len was saying about INT 9 being paired with INT 2, and
therefore it needs watching ???.
Interrupts 8-15 are latched through the second of two interrupt
controllers (0-7 are on the first one). The design of the original PC
only had one interrupt controller and it would interrupt the CPU (808?)
whenever any one of the 8 interrupts were triggered. With the advent of
the second controller, it was decided to keep the original design, so
now if interrupts 8-15 occur, they are latched by the second controller,
and irq 2 is latched by the first controller, and then the CPU is
interrupted. The interrupt service routine (ISR) knows all about this
behaviour, and can deduce which IRQ (0-15) really is responsible for the
interrupt.

In short, IRQ 9 is fine.