tulip interacting with cyclades?
Jon Lewis
jlewis@inorganic5.fdt.net
Fri Nov 13 23:59:23 1998
This may be borderline off-topic, but the following forwarded message is a
reply from Marcio of Cyclades to me earlier today (yes...I asked his
permission to forward it here :)
The short story is, I've got this P100 dialup server (64 ports via
cyclades 16ye cards) that can't be warm booted if it has a Netgear FA310TX
in it (the old 21140 kind). For the longer story, read below. For
maximum clarity, page down to the bottom and read the last message first.
This looks to me like either some kind of hardware problem (or more
likely, bad interactions between 2 drivers or 1 driver and what should be
unrelated hardware). Anyone have an idea of what could be going on?
---dont't waste your cpu, crack rc5...www.distributed.net team enzo---
Jon Lewis <jlewis@fdt.net> | Spammers will be winnuked or
Network Administrator | nestea'd...whatever it takes
Florida Digital Turnpike | to get the job done.
______http://inorganic5.fdt.net/~jlewis/pgp for PGP public key________
---------- Forwarded message ----------
Date: Fri Nov 13 23:59:23 1998
From: Marcio Saito <marcio@cyclades.com>
To: Jon Lewis <jlewis@gnv.fdt.net>
Cc: Doris Li <doris@cyclades.com>, Ivan Passos <ivan@cyclades.com>
Subject: Re: cyclades reboot problem solved
Resent-Date: Fri, 13 Nov 1998 23:52:05 -0500 (EST)
Resent-From: Jon Lewis <jlewis@gnv.fdt.net>
Resent-To: jlewis@inorganic5.fdt.net
Resent-Subject: Re: cyclades reboot problem solved
Jon:
I don't remember you reporting the reboot problem. In any case, I
understand the problem. If you perform a "warm reboot", the system
won't come up if the Cyclom-Ye and the NetGear Ethernet are installed
in the same system. Everything works fine if the system is powered off
and then on again.
It is my understanding that the NetGear Board uses one IRQ, some I/O
addresses and no Memory addresses. The Cyclom-Y/ISA board uses one
IRQ, memory addresses and no I/O addresses.
The IRQs are different (5 for the Cyclom-Y and 14 for the NetGear).
So, unless there is something very strange hapenning in hardware,
there should be no reason for any type of conflict.
Based on that, my conclusion is that the problem is related to the
drivers (maybe one driver is interfering with the other's hardware or
vice-versa during driver init time). The question is: Why the system
comes up ok after a power-cycle and not after a warm reboot?
Hardware-wise, the only different is that the is no HW reset in the
PCI bus in a warm reboot.
I thought about that and couldn't get to any conclusive answer.
I don't know exactly how you are loading the modules, but is it
possible that the order in which the two drivers are loaded is
different in a warm and cold reboot? If that is the case, I would try
to reverse the order in which the drivers are loaded and see what
happens (maybe nothing changes. maybe everything works. maybe the
system locks everytime).
Also, I would try to identify whether is the physical presence of the
boards in the bus that cause the system lockup or it happens only when
both drivers are also present.
If you have the opportunity to try or verify any of that, please let
me know.
Regards,
Marcio.
Jon Lewis wrote:
>
> I don't know if you remember the cyclades reboot problem I mentioned a few
> months ago...but ever since we moved to our new building, I was unable to
> do a "shutdown -rf now" on endor (our P100, dual 16ye, 64-port dialup
> server). Doing so, resulted in the kernel hanging when the cyclades
> module loaded. The only way to reboot was to power cycle the system.
>
> This morning, it locked up, so I had to go in early, and decided to open
> it up to make sure the CPU fan was working (it was) and also decided to
> swap ethernet cards. When we moved (last May), we upgraded to a 100mb
> ethernet switch, and I upgraded many of our systems to Netgear FA310TX
> fast ethernet cards. This seems to have caused the problem. Swapping out
> the netgear for an old 3c509 solved the reboot problem.
>
> I really have no idea why the Netgear card should cause problems with the
> Cyclades cards...but that's the conclusion I've come to. Here's how
> things were/are set.
>
> endor:~# cat /proc/interrupts
> 0: 647423 timer
> 1: 64 keyboard
> 2: 0 cascade
> 5: 6644401 + cyclades
> 8: 0 + rtc
> 9: 23261 + ncr53c8xx
> 10: 228050 3c509
> 11: 3996442 + cyclades
> 13: 1 math error
>
> When the Netgear was there, it was on IRQ 14 (on-board IDE is disabled).
> Cyclades card configs (these are the ISA ones) have not changed for at
> least a year or so.
>
> Cyclom driver $Revision: 1.36.3.9 $$Date: 1996/10/07 19:47:13 $
> Cyclom-Y/ISA #1: 0xd4000-0xd5fff, IRQ11, 32 channels starting from port 0.
> Cyclom-Y/ISA #2: 0xd6000-0xd7fff, IRQ5, 32 channels starting from port 32.
> eth0: 3c509 at 0x300 tag 1, 10baseT port, address 00 a0 24 04 a4 c8, IRQ 10.
> 3c509.c:1.12 6/4/97 becker@cesdis.gsfc.nasa.gov
>
> Here are the message from the Netgear when it was there:
>
> tulip.c:v0.88 4/7/98 becker@cesdis.gsfc.nasa.gov
> Digital DS21140 Tulip at 0x6100, 00 40 05 41 8e 68, IRQ 14.
> eth0: EEPROM default media type Autosense.
> eth0: Index #0 - Media MII (#11) described by a 21140 MII PHY (1) block.
> eth0: MII transceiver found at MDIO address 0, config 1000 status 782d.
> eth0: Advertising 01e1 on PHY 0, previously advertising 01e1.
> eth0: Advertising 01e1 on PHY 0 (0).
> eth0: Setting full-duplex based on MII Xcvr #0 parter capability of 01e1.
>
> Any ideas? Our rocketport based system works fine with a netgear card.
>
> ------------------------------------------------------------------
> Jon Lewis <jlewis@fdt.net> | Unsolicited commercial e-mail will
> Network Administrator | be proof-read for $199/message.
> Florida Digital Turnpike |
> ______http://inorganic5.fdt.net/~jlewis/pgp for PGP public key____
--
_________________________________________________________
Marcio Saito
Product Manager/Director of Technology
voice (510) 770-9727 x214 mailto:marcio@cyclades.com
Cyclades Corporation http://www.cyclades.com
_________________________________________________________