[479] in linux-net channel archive
Re: PLIP still a no go
daemon@ATHENA.MIT.EDU (Avery Pennarun)
Tue Jun 13 18:27:35 1995
Date: Tue, 13 Jun 1995 17:05:48 -0400 (EDT)
From: Avery Pennarun <apenwarr@foxnet.net>
To: Ken Estes <m-ke0082@SPARKY.CS.NYU.EDU>
cc: linux-net@vger.rutgers.edu
In-Reply-To: <9506130357.AA08798@SPARKY.CS.NYU.EDU>
On Mon, 12 Jun 1995, Ken Estes wrote:
> I often have to add the route to the
> other computer back to the route table because it gets dropped. I
> have also added default gw other.host to the route to help ensure the
> packets get through (this has helped me on the real Internet with
> routing problems I have never been able to track down).
Kill routed. It's only useful in special situations, and even then it only
works sometimes. :) That's what's eating your routes.
> One suggestion read:
> >Make sure the other system (which is likely to be set to port 0x378) has irq
> >7 free, and make sure your I/O card has parallel port irq's enabled and set
> >to the right irq! My I/O card, being stupid, came configured for 0x378 and
> >IRQ 5. To top that off, the manual lied and said these settings REALLY
> >meant IRQ 7. Sure, sure. Either way, make sure it's set correctly :)
> > [...]
>
> Go a bit slower here, how do I check to ensure the port is free? I
> only have a disk drive in the machine (not even a graphics card is
> plugged in, I am using the built in vga port on the back. It is an old
> machine) I have the cable plugged into the parallel port that came
> with the machine. I have done nothing special to the machine besides
> put linux on the disk.
What I meant was, your two plip drivers are set to different addresses:
plip0 0x3bc irq5
plip1 0x378 irq7
At least, this is what Linux thinks you have. If the port addresses were
the same, both devices would be the same (either plip0 or plip1). This is
to answer your question as to why the device names are different.
Now, Linux-plip assumes a port set to 0x378 is going to be on irq7. This
may or may not be true - and as I said above, my card came (somehow)
misconfigured to 0x378 and irq5, while the MANUAL for the card said the
equivalent DIP switch settings were 0x378 and irq7, leading to immense piles
of confusion.
My results were _exactly_ the same as your ifconfigs seem to indicate -
packets transferred perfectly one way, responses were sent, but were never
received. This is (at least in my case) because of an irq conflict on the
bad receiver.
(Actually, it was even worse for me since I already had a device on IRQ5!)
So, anyway, check to make sure there are no other devices in your computer
using irq5 or irq7 (from what you said above, probably there aren't) and
then make sure the parallel card is set to the same irq as the linux driver.
The best way to make sure the cable is right is to boot dos 6.x on the two
machines and try to run interlnk/intersvr between them. Since interlink
doesn't use IRQ's, it will make sure all the hardware other than irq's is
working correctly.
If this test works, irq's are definitely your problem. Try moving the
parallel port address/irq jumpers on the card, and try both irq5 and irq7 in
the linux plip driver.
It's also quite possible that the I/O card doesn't do parallel port
interrupts at all - this is bad, and the only way to fix it is to get a new
parallel port card. This is a last resort, of course, but I have heard of
cards that can't do parallel port irq's.
There, hopefully that made a bit more sense :)
Good luck!
(P.S. Nimbus is the one that is likely to have the irq conflict, since it
can't receive: cirrus is working fine.)
