SCO 3.2v4.2 - How to resolve an address conflict?
Volker Raab
we have a severe problem with our server and don't know how to solve it.
Our problem is that the network is not functioning. We get the following
error when TCP/IP starts:
slink: Funktion "cenetb", command 1: open "/dev/emG0" no such device or
address
The best state we reached was that the error appeared a bit later (in the
startup-script /etc/rc2.d/S85tcp) when running the command
ifconfig: ioctl (SIOCGIFFLAGS): no such interface
But: At boottime we see that the driver is successfully initialized, because
the chip and the MAC-address are pronounced correctly.
hwconfig lists the following lines:
%adapter 0xE000-0xE060 15 0 type=bhba unit=0 id=7 Chip=53c875-E
%emG0 0xDC00-0xE400 11 - type=DC21040/BNC addr=000092B61397
The MAC address could be read, so the NIC initialized.
But have a look at the adreesses of the SCSI-adapter and the NIC!
The address space of the SCSI lies completely in the address space of the
NIC, so that is - we guess - why the NIC is not functioning.
It is a Cogent eMASTER+ PCI (EM960) with LLI Rev. 1.0.5
The SCSI card is also PCI SC875. We tried to upgrade the Firmware BIOS an
were successful to v4.16.00 while the SCO-Drivers are 3.06.00, but we could
not change the address of the SCSI-card.
Our system is SCO OpenDesktop 3 (3.2v4.2) and we have also installed SCO
UNIX Development system Release 3.2v4.2
Is it possible to change the address space of the Cogent NIC and compile the
driver. Source code seems to be included (there is a space.c file).
Or is there an other possibility to resolve the obvious address-conflict?
If the problem cannot be resolved maybe there are some people who have this
old but good system still running with network up and can tell us WHAT NIC
to use and WHERE to get the LLI-drivers for it.
Maybe we can find or by another card anywhere and solve the problem by
changing the hardware.
But remember: We do not have any ISA-slots. Only PCI!!!!
Help or hints would be greatly apprecheated
TIA,
R. Bahcic
Abid Khan
I have only used ISA cards on SCO V/386 3.2.4, but take a look at 3com
model # 3C590-TPO 3Com® EtherLink® III Parallel Tasking PCI Bus Master
Network Interface Card
here is the link:
http://www.3com.com/products/en_US/searchbyfile.jsp?path=download&sort=prodnum&fileid=5&prodcat=19&fname=&os=22913
Hope this helps
Abid
Bela Lubkin
> Dear SCO 3.2v4.2 Nostalgics,
>
> we have a severe problem with our server and don't know how to solve it.
>
> Our problem is that the network is not functioning. We get the following
> error when TCP/IP starts:
>
> slink: Funktion "cenetb", command 1: open "/dev/emG0" no such device or
> address
>
> The best state we reached was that the error appeared a bit later (in the
> startup-script /etc/rc2.d/S85tcp) when running the command
>
> ifconfig: ioctl (SIOCGIFFLAGS): no such interface
>
> But: At boottime we see that the driver is successfully initialized, because
> the chip and the MAC-address are pronounced correctly.
>
> hwconfig lists the following lines:
>
> %adapter 0xE000-0xE060 15 0 type=bhba unit=0 id=7 Chip=53c875-E
> %emG0 0xDC00-0xE400 11 - type=DC21040/BNC addr=000092B61397
>
> The MAC address could be read, so the NIC initialized.
> But have a look at the adreesses of the SCSI-adapter and the NIC!
> The address space of the SCSI lies completely in the address space of the
> NIC, so that is - we guess - why the NIC is not functioning.
It's a good guess. It looks like the machine's BIOS has mis-assigned
PCI I/O space. The BIOS is supposed to do this, and it isn't supposed
to leave any devices overlapping each other. The addresses chosen by
the BIOS are typically influenced by what slot each board is in;
rearranging slots might fix the problem.
> Is it possible to change the address space of the Cogent NIC and compile the
> driver. Source code seems to be included (there is a space.c file).
> Or is there an other possibility to resolve the obvious address-conflict?
The drives should automatically adjust to moved boards. Or, you might
have to run `netconfig` and tell it about the NIC's new slot number.
>Bela<
Jeff Liebermann
wrote:
>It's a good guess. It looks like the machine's BIOS has mis-assigned
>PCI I/O space. The BIOS is supposed to do this, and it isn't supposed
>to leave any devices overlapping each other. The addresses chosen by
>the BIOS are typically influenced by what slot each board is in;
>rearranging slots might fix the problem.
>
>> Is it possible to change the address space of the Cogent NIC and compile the
>> driver. Source code seems to be included (there is a space.c file).
>> Or is there an other possibility to resolve the obvious address-conflict?
>
>The drives should automatically adjust to moved boards. Or, you might
>have to run `netconfig` and tell it about the NIC's new slot number.
>>Bela<
Most BIOS's have a "Reset System Configuration Data" or "Reset
Extended Configuration Data" option in the CMOS setup. It's always
"DISABLED" by default. It's primary purpose is to recover from memory
or port allocation errors that leave the machine unbootable. It will
also fix overlapping memory map assignments. Set it to "Enabled",
save the settings, and power down the machine. That means pull the
plug and let it sit for a while so that the "wake on lan" feature
turns off the etherknot card. When it boots, the bios will announce
that the ESCD has been reset. With any luck, it should then work. No
changes needed to OSR5.
--
Jeff Liebermann 150 Felker St #D Santa Cruz CA 95060
(831)421-6491 pgr (831)336-2558 home
http://www.LearnByDestroying.com WB6SSY
je...@comix.santa-cruz.ca.us je...@cruzio.com
Volker Raab
We could reassign the addresses of the PCI-devices by rearranging them in
the slots as you told us.
But we are sad to tell you that the problem still is not solved.
there comes still the message
slink: Funktion "cenetb", command 1: open "/dev/emG0" no such device or
address
despite the fact that all seems to be configured correctly.
/dev/emG0 has 40,79 Major,Minor numbers and 40 is in mdevice "clone" while
the major-device of the emG0 is 79.
could there be the error?
Shouldn't be the devices Major/Minor combination begin with 79???
Hope your still able to help us.
Thank you once more.
TIA,
R. Bahcic
Jeff Liebermann <je...@comix.santa-cruz.ca.us> schrieb in im Newsbeitrag:
mmjk0v8igd8h5g367...@4ax.com...
Bela Lubkin
> We could reassign the addresses of the PCI-devices by rearranging them in
> the slots as you told us.
>
> But we are sad to tell you that the problem still is not solved.
>
> there comes still the message
> slink: Funktion "cenetb", command 1: open "/dev/emG0" no such device or
> address
> despite the fact that all seems to be configured correctly.
>
> /dev/emG0 has 40,79 Major,Minor numbers and 40 is in mdevice "clone" while
> the major-device of the emG0 is 79.
>
> could there be the error?
> Shouldn't be the devices Major/Minor combination begin with 79???
No, that's how clone drivers work. The device node of any clone driver
should have major = clone, minor = device's major.
``"/dev/emG0" no such device or address'' means that the hardware is not
being recognized by the driver. After moving the card, did you run
`netconfig` and tell it the card's new slot number? (It probably wants
three numbers labeled "bus, device, function". The "device" number is
really the slot number.)
>Bela<
Volker Raab
Bela Lubkin <be...@caldera.com> schrieb in im Newsbeitrag:
2002122615...@mammoth.ca.caldera.com...
Yes of course, we ran netconfig and added a chain sco_tcp -> emG0
But we were not asked for a slot number (!) only what media type (TP/BNC or
Autosense)
And the device initialises correctly.
"hwconfig" tells us the right (!) MAC-Address of the card at boot time.
Any other idea?
Thank you so much to the moment, please help further.
TIA,
R. Bahcic