• Welcome to OS2World OLD-STATIC-BACKUP Forum.
 

News:

This is an old OS2World backup forum for reference only. IT IS READ ONLY!!!

If you need help with OS/2 - eComStation visit http://www.os2world.com/forum

Main Menu

Networking configuration issue - boot does not complete

Started by Peter400, 2010.07.29, 02:32:59

Previous topic - Next topic

RobertM

IO Address 320 may conflict with various sound cards, if memory serves. I (back in the day) avoided (like the plague) anything between 300 and 340. Disabling any MIDI support (or, assuming it disables MIDI and other sound functions; disabling sound support) may prevent the issues, if it's similar to the ones I had. It has been a very long time, but I seem to remember it overlapping MIDI support (SB and compatibles and standard "back in the day" MIDI) or control support (some ESS/CS/mWave implementations) for various sound devices.

IF I recall correctly, the base address for MIDI/Game was usually 300, 330, or (configured to) something inbetween (such as 320) usually in increments of 10 (ie: 300/310/320/330). If memory serves, it's usually 330 and otherwise 300 by default on most of the common sound cards, but I have found other variants (310, 320, 500s, and so on).

hth,
Rob


|
|
Kirk's 5 Year Mission Continues at:
Star Trek New Voyages
|
|


Peter400

Quote from: RobertM on 2010.08.25, 23:35:32
IF I recall correctly, the base address for MIDI/Game was usually 300, 330, or (configured to) something inbetween (such as 320) usually in increments of 10 (ie: 300/310/320/330). If memory serves, it's usually 330 and otherwise 300 by default on most of the common sound cards, but I have found other variants (310, 320, 500s, and so on).

hth,
Rob

I double checked, and there's no conflict. I do have a sound card, but it uses some addresses in the 500s range.
Nevertheless I tried to use 2C0 for a change, but the result is identical - The driver talks to the card, but all transmissions fail.

Peter

Peter400

Quote from: warpcafe on 2010.08.25, 14:12:58
well, there still is this one:
http://os2warp.be/index2.php?name=nicpak

It's a link to Chuck McKinnis' Nicpak which also lists a zip file for PCMCIA cards (an the ZIP can be downloaded).
This ZIP from Chuck has drivers for various Xircom pccards (listed on the page).
While Nicpak installer cannot detect PCMCIA cards and thus install drivers automatically, it at least HAS drivers in case we need. ;)
(Just as a last resort in case there is someting fishy about Peters drivers...)

I downloaded this package + pcmcia.zip and I see that the driver contained there is the same I'm already using.
But speaking of Chuck McKinnis, here's an old note from him on Xircom card success:

http://www.os2voice.org/VNL/past_issues/VNL0701H/vnewsft.htm

Unfortunately, this does not work on my laptop >:(

Peter

Andi

Quote from: Peter400 on 2010.08.25, 21:06:12
Hi Andi,

Quote from: Andi on 2010.08.25, 11:31:42
Quote from: Peter400 on 2010.08.25, 04:49:51
....
I also tried it without enforcing IO mode, and without enforcing line speed - no success.
I'm using 10 Mbit setting, because the card is said to work in full duplex mode at this speed. Only half-duplex is possible with 100 Mbit.

Peter

Usually forcing Link speed/mode is a bad thing. It is guaranteed to work ONLY when the link partner is forced to the same settings too. Of course most times it will work if the other one is in Auto mode, but not for sure.

Your are sure your cabling is okay? Have you tried this setup with other OSes?

Yes, my router should handle 10 Mbit Ethernet. And another reason I went with 10 Mbit is that the Xircom dongle has a little light which is supposed to go on when the card works in 10 Mbit mode. The light goes on, so it provides me yet another assurance that the driver actually talks to the card.
You missed my point here. I'm pretty sure your router is able to work in 10Base-T mode. But if your router is set up to 'Auto negotiate', which I think every 100Base-TX capable device per default is, it's not guaranteed that it will link to any 'forced' device. Point is, don't force any device or force both devices. Otherwise it's not guaranteed to work.

When the LED lits up this usually means the PHY has established a link. As PHYs usually do not need any configuration, this may indicate something driver related or not. It may indicates only that the PCMCIA card is powered up. But that's a guess.

Although most times it works, sometimes even with link established (LED lits)  no data can travel over the connection. I saw this especially with bad or too long cables. That's why I asked for your cabling.

Quote
The setup works without any changes in Linux and Windows 98.
Linux says on startup (dmesg):
   eth0: Xircom: port 0x300, irq 3, hwaddr <MAC-address-here>

In Windows I see the following resources:
  Input/Output Range   1020-102F
  Interrupt Request  03
  Memory Range    0A000000 - 0A000FFF
  Memory Range    0A001000 - 0A001FFF

In case you wonder, yes, I tried using IRQ 3 as well. It made no difference.

Peter

You suppose the driver can set up IO-Addresses and IRQs as in another post you wrote you'll try to use addresses starting at 320 and IRQ 10 which is different to what you use in W98. I've never seen a PCMCIA network card/driver which can do this. That does not mean that it is impossible. But in my experience it is not very likely. Or maybe I'm only used to older cards/drivers which definitely can not do such things.

Peter400

Quote from: Andi on 2010.08.26, 23:53:43
You suppose the driver can set up IO-Addresses and IRQs as in another post you wrote you'll try to use addresses starting at 320 and IRQ 10 which is different to what you use in W98. I've never seen a PCMCIA network card/driver which can do this. That does not mean that it is impossible. But in my experience it is not very likely. Or maybe I'm only used to older cards/drivers which definitely can not do such things.

Yes, you are right. I expect the driver to assign IO address and IRQ as configured. (Whether the card will work after that is a completely different matter.)
If it was not the case, the card would have a hard-coded IRQ of 3. I find this unlikely, because by default this IRQ had been used by serial communication ports for decades. It just happens that this is the lowest unused IRQ on my laptop, so both Windows and Linux grab and use it for the card.

Also, if I do not specify any IRQ, the driver says "A working IRQ could not be found" and then "The network is running in polled mode" before the boot process freezes.

Furthermore, the .NIF file description for the IRQ setting lists this piece of help: "This parameter selects a hardware interrupt for use by the adapter.", suggesting that the driver is in control of which IRQ will be used.

Peter


Peter400

Hi folks,

I was not able to get to the bootom of the original problem and solve it. I still think there's something wrong with the interrupts from the PC Cards - they do not reach the driver.

Anyway, I found another Xircom card: RealPort CardBus Ethernet 10/100 + Modem 56. This is a newer and better card (the other ones I tried were just PCMCIA cards, not CardBus), and the driver is newer, too. And the best part is that it all works! DHCP works, too! (well, most of the time...)
So I no longer need to solve the PC Card problem ...

I did not have to change any settings. I'm not using any socket and card services (as instructed by Xircom).

Interestingly, while the driver loads, it displays the following message:

Xircom CardBus Ethernet 10/100 Adapter NDIS v3.14 (010612)
...
PCI IRQ routing unavailable - the driver is using ISA IRQ routing
Memory address 5B800000, IO Address F500, IRQ 11, Node Address <node-addr>


I do not really understand the IRQ routing secrets, but the message above seems to confirm that there's probably something unusual with my laptop's CardBus controller TI-1225, something that confuses OS/2 only, because Windows and Linux did not have a problem.

Once again thanks for everyone's help

Peter