• 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

SCSI ASC.ADD

Started by Raiko, 2008.10.17, 17:26:57

Previous topic - Next topic

ddan

Newer systems get sticky as to hardware interrupts, which may be impossible to resolve -- with given hardware, because so many ports are hard-wired -- so unless you've tested on the exact same hardware, that's where I'm leaning toward as the essential cause. You can use PCI.EXE to look at the interrupts and see if two or more are shared, always a bad sign for OS/2. If you get down to this, you can try turning off unused COM ports in BIOS, in fact just for a trial turn off everything that can be, to change how the interrupts are assigned. Also, think that OS/2 never uses an interrupt for VGA cards, so turn it off -- if possible. All hardware not strictly necessary for booting turned off (or removed) would likely give an indication as to whether interrupts are the root cause.

Raiko

When remming out acpi.psd the driver loads without any errors.
(With acpi I've also tried remapping irq's and various parameters.)

ddan

Okay, I'm at the limit of my knowledge, but clearly interrupt assignment is the problem, and looks to me as though you MUST free up at least one that the card can use. Or, use PCI.EXE and with its list try more re-arranging. I'm assuming you tried swapping slots as Warpcafe suggested.

Saijin_Naib

There could be a fun ACPI issue with the version of ACPI that you are testing with. I know that some versions will break/fix compatibility with many different chipsets and drivers and you may need to revert to an old version or try the latest testcases. I've been fortunate in that every release has worked with my VIA chipset but I've read that others have had intermittent issues.

Have you tried mucking about with ACPI switches on the driver further? I don't see any mention of trying the /CD (crazy devices) switch which may or may not do anything.

Raiko

Running pci.exe gives an error:

ROM PCI IRQ routing table Tests....
ROM IRQ routing table found at F000h:DE80h
Table Version 1.0 - OK
Table size 352 bytes - OK
Table Checksum 00h - Failed!
IRQ's dedicated to PCI : None
The ROM PCI IRQ routing table appears to be faulty!!

IRQ Summary: IRQs 3,11,16,17,18,22 are used by PCI devices
Shared IRQs: IRQ 16 is shared by 7 PCI Devices
             IRQ 17 is shared by 5 PCI Devices
             IRQ 18 is shared by 5 PCI Devices

(The SCSI adapter uses IRQ3)

I have also tried a testcase acpi, 3.09 and 3.11. With and w/o /CD and also swapping cards.

Like I mentioned before with /SMP /APIC /!NOD all drivers load without any errors, but the system
freezes just before starting the WPS.
How to identify what causes this ?


Saijin_Naib

What display driver are you using? Panorama or SNAP or VGA? Another driver may not be ACPI ready.

Raiko

I'm using Panorama

Pete

Hi

The system freeze (? - do you mean hang?) at WPS startup could be audio related. Switch off Audio in the BIOS to test.

I seem to recall the Advansys controller preferring IRQ9; I think the only other IRQ I had 1 of those working on was 10. I'm a bit surprised to see it on 3...

As the controller works without ACPI installed the problem seems to be ACPI related. No idea if the ACPI-SCSI.doc is of any help but you should have a copy with whichever version of ACPI.

Some versions of ACPI need the /TMR switch here (you can test that if you can boot to a working system with acpi installed) by running the TMRTEST.EXE app. Silly results mean you need that switch to produce output that looks like this:-

[L:\ecs\bin]tmrtest
rc=0  Resolution:1193167
Elapsed 0 sec. Must be 0
Elapsed 2 sec. Must be 2
Elapsed 4 sec. Must be 4
Elapsed 6 sec. Must be 6
Elapsed 8 sec. Must be 8
Elapsed 10 sec. Must be 10
Elapsed 12 sec. Must be 12
.
.
.


Regards

Pete



ddan

Since most of the interrupt assignment is done by BIOS, you may be able to shorten the try/look cycle by booting DOS from floppy, then running PCI.EXE and checking the table info. I did that (successfully) at least once.

Raiko

Even if I had a DOS floppy I doubt that I would be able to fix any problem it
might display. This SCSi card is very old and I don't think it is worth the trouble
to continue trying with ACPI.
I'd better get a newer adapter so if anyone knows about one that works well
with ACPI please let me know.

RobertM

Quote from: Raiko on 2008.10.17, 17:26:57
I'm using asc.add for an Advansys ABP-915 and the driver displays
the following during boot:
AscInitGetConfig() warn_code 128
Does anyone know what it means?

Hi Raiko,

Failing being able to go into the SCSI BIOS (sorry, not familiar with that particular card) and re-assigning IRQs and/or address space, I would suggest a newer Adaptec card (first, ensure it is supported by the drivers available). And if it is an IBM system, you may wish to check into the various IBM ServeRAID cards. Not all are supported, but there is a cross reference chart online (if you cant find it, let me know and I will dig it up).

If you can go into the SCSI BIOS (usually there is a message on boot similar to "Hit CTRL-F to enter SCSI BIOS/SETUP/OPTION/ETC", try changing either the IRQ or the card's base address. On certain cards (again, dont know if this will apply, as I am not familiar with that card), if you change the card's base address, you may have to add a switch to the driver parameter in the config.sys file... something like /B:0F00 or similar to tell the driver where the card is "located"




Perhaps a little more about what you are trying to accomplish and what types of drives and devices you are connecting and planning to in the future would be helpful - we would be able to recommend a particular SCSI card based off those needs.

-Robert


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


Raiko

I intended to use a scanner with this card.
The driver loads, and works, without ACPI enabled.
Tried lots of parms with ACPI, but no success. Anyway, the scanner also works with USB
so I don't really need the SCSI card. (Don't think this old card is much used so I won't
bother the ACPI-guys with it.)

rwklein

What Thomas wrote:

Hi,

most probably and according to Google search like
http://orange.kame.net/dev/cvsweb.cgi/kame/netbsd/sys/dev/ic/adwlib.h?rev=1.1
(serach for "warn_code" in the page) it stands for
/* PCI config space set error */
...which can pretty much mean everything related to PnP I'm afraid. Like wrong IRQ, shared IRQ, whatever.
Try plugging the SCSI controller card into a different PCI slot (if possible) in that case.

Perhaps it also means
/* set SCSI ID failed */
...which in turn can be related to wrong/duplicate SCSI-adressing of a device. Perhaps also a mis-configuration of SCSI BIOS options (if the controller has a configurable BIOS).

Hope this helps (although not sure as the above is drawn from Linux driver code and heavily "interpreted"...)
Thomas"

Then try and rem out ACPI this might do something wrong in the PCI config.sys.
If it then works then open a ticket on Netlabs track for ACPI.

http://svn.netlabs.org/acpi

Roderick Klein
Mensys