VOGONS


First post, by dionb

User metadata
Rank l33t++
Rank
l33t++

Getting a SCSI CD-Rom working under DOS should be simple:

- add adapter card into system
- hook up drive with known-good cable and termination
- select free SCSI ID
- get correct ASPI driver for the card and load in CONFIG.SYS
- load ASPICD.SYS
- load MSCDEX.EXE
- profit.

However I'm getting stuck at step four. I have a known-good AHA-2940AU (works in Win98SE in different system), known good Plextor PX-40TS CD drive set to SCSI ID 3 and a known-good 50p cable connecting both.

At POST the SCSI BIOS is active and detects CD drive at ID3.

Then CONFIG.SYS gets to DEVICEHIGH=ASPI8DOS.SYS /D (should be correct for AHA-2940AU). It correctly detects the AHA-2940AU:

Host adapter 0
SCSI ID 7
Interrupt level 7
PCI ID 0:2:0

Then it hangs for a few minutes and after timeout gives:
"aspi8dos.sys installation failed"

Of course ASPICD.SYS fails too because no ASPI driver installed "no valid Adaptec host adapter".

There's more going on as if I attach a HDD, FDISK hangs too on being run, then dropping to command prompt. First thought was an IRQ conflict, but I have LPT1 disabled and nothing else on IRQ7. MSD confirms nothing detected there.

Rest of system:
Cyrix Cx5x86-100
8MB RAM (default: HIMEM.SYS and EMM386.EXE RAM 4096, but also tried and failed with no memory management at all)
UMC UM8881F -based PCI So3 motherboard
Diamond Stealth 3D 2000 (Virge PCI) VGA
All other cards removed to rule out conflicts (previously: 3C509C on IRQ 10 and ALS100 SB16 clone on IRQ 5 and DMA 1+5, no change in SCSI symptoms after removing)
2GB IDE DOM

So: can anyone offer any pointers on why known-good components aren't playing ball under DOS on a system with no visible confilcts?

Btw, endgame is to use a RaSCSI emulated SCSI CDRom drive, but before I get on to exotic stuff, I want to be sure the basics are working with very simple parts.

Reply 1 of 5, by Grzyb

User metadata
Rank Oldbie
Rank
Oldbie

Some Adaptec drivers don't like to be loaded high.

Nie tylko, jak widzicie, w tym trudność, że nie zdołacie wejść na moją górę, lecz i w tym, że ja do was cały zejść nie mogę, gdyż schodząc, gubię po drodze to, co miałem donieść.

Reply 2 of 5, by dionb

User metadata
Rank l33t++
Rank
l33t++
Grzyb wrote on 2020-12-08, 08:18:

Some Adaptec drivers don't like to be loaded high.

Should have added: when I disabled HIMEM.SYS and EMM386.EXE I also changed DEVICEHIGH into DEVICE. No change unfortunately.

Reply 4 of 5, by dionb

User metadata
Rank l33t++
Rank
l33t++

Tried another slot: PCI ID 0:1:0 - but still hangs/timeouts/doesn't install.
Forced IRQ 11 and 10 (BIOS lets me set first/second/third/fourth PCI/PnP IRQ, this card always gets first): IRQ 11 or 10 detected - but still hangs/timeouts/doesn't install.
Changed PCI IRQ trigger from EDGE to LEVEL: no change, still hangs/timeouts/doesn't install

But: this BIOS has lots of PCI options. It also allows assigning INT line per slot to A, B, C, D or AUTO. Default is slot 1 A, slot 2 B, Slot 3 C.

On a hunch I set them all to AUTO. It worked! The card (in slot 2 PCI ID 0:1:0) received IRQ11 (2nd on list) and ASPI8DOS.SYS loaded without a hitch.

Then tried to manually reproduce what was going on. Turns out INT A works, regardless of slot. B and C do not. It definitely is the interrupt line, not the IRQ, as changing order of IRQs changes the IRQ it gets - but does not change whether it works or not. IRQ 10 or 11 + INT A works, same IRQ + INT B does not. Or if in doubt: just set it all to AUTO and it works too.

So, success!

Thanks for pointers in the right direction 😀

For future reference (as this seems more motherboard related than SCSI-specific) it's an ADI TK8881 FA20 motherboard, with "UMC 486 Mainboard" sticker helpfully plastered over the model name.

Edit: and for the record, ASPI8DOS.SYS and ASPICD.SYS are perfectly happy to be loaded high. freeing up over 37k of conventional memory in the process.