Adaptec SCSI card DOS driver woes (mini-rant)

[Stinkerton18]

This is a bit more of a mini-rant at the rather abhorrent state and experiences I've had with the real mode drivers for Adaptec's older, non-PCI cards. Comments/suggestions are of course welcomed, more wanting to share my experiences in hopes that maybe I can spare someone else this same pain.

Let's start out with the system in question's specs:

• Super Micro P54VLPCI Motherboard
  • Socket 5, 4 72-pin SIMM sockets, 2 VLB, 3 PCI, 2 ISA (excluding the VLB ones), 1 ISA/PCI
  • Pentium 90 CPU
  • Trident ISA video card
  • Adaptec 1542CF ISA SCSI card (Floppy port enabled and connected)
  • Adaptec 29160N PCI SCSI card
  • Plextor 8x SCSI CDROM drive, Caddy load (shocked me too)
  • ISA Super I/O card (floppy, IDE, and Serial Port 2 all disabled)
  • Linksys LNE100TX PCI NIC

The primary use for this adorable tower is mainly for doing DOS tool testing and config vetting, and help the BlueSCSI/PiSCSI projects by making "ready to run" DOS 6.22/Win98 DOS drive images that have ASPI CD/removable media drivers already setup. This way I don't have to worry about breaking my own DOS gaming machine by accident.

SCSI Drives connected externally (in order) on 1542CF:

• 128MB Magneto Optical Drive (termination disabled), ID 5
• BlueSCSI v1 with:
  • 512MB HD image, ID 0
  • ISO file used as CDROM, ID 6

SCSI Drives connected (in order) on 29160N

• Plextor 8X CDROM (internal, on 50-pin Fast SCSI connector)

On POST, the1542CF card sees all three drives, and loads its bios as expected. The 29160N card sees the Plextor, but does not load its bios because it sees no hard disks/removable media drives (CDROMs don't count here, only the Magneto Optical, Zip, Jazz, etc.)

Now the image I had made previously does work, which already had the ASPI8DOS.SYS driver (for the majority of Adaptec's PCI based cards). Those load just fine, the ASPICD.SYS driver loads fine, and after MSCDEX/SHSUCDX loads, the Plextor is given a drive letter and perfectly acessible. Thanks to enabling the option to "treat removable disks as fixed" enabled on the 1542CF, the magneto optical drive is fully usable in DOS, and the system boots completely as expected on the hard drive image.

My woes start when I try to load ASPI4DOS.SYS, the ASPI driver for the 1540 series and 1640 Microchannel cards. The driver DOES see the 1542CF, but it just....hangs until it times out reporting it can't load. Subsequently the ASPICD.SYS driver fails to load, then things go way off the rails and even the hard drive stops responding/system halts.

I *suspect* it's an IRQ conflict, but still need to test more.

 

[Stinkerton18]

After the holidays, and the loss of a family member, I'm finally home and could pick this back up. SOOOO, strap in cause it gets a little weird.

First thing I did was move the AHA-1542CF, along with the ISA Super I/O card. From top (closest to the keyboard connector) to bottom, here is how the cards are now arranged:

PCI slot 1
PCI slot 2	Linksys LNE100TX PCI NIC
PCI slot 3	Adaptec 29160N PCI SCSI card
PCI Slot 4
PCI/ISA combo Slot
VLB/ISA Slot 1	Adaptec 1542CF ISA SCSI card
VLB/ISA Slot 2	ISA Super I/O card
ISA Slot	Trident ISA video card

Next up, moving IRQs around a bit as I still suspect IRQ conflicts.
The 1542CF is set to use IRQ 9 (set in the BIOS config utility of the card itself), DMA 5
The 29160N is set to use IRQ 11

Both cards are set for the default I/O address range

Now with these settings, the ASPI8DOS.SYS driver for the Ultra2 card works fine, but sadly ASPI4DOS.SYS still does not load. On a whim, I pulled the PCI 29160 card out entirely. No matter what ISA slot the 1542CF is in, it just does not work, even when it and the Trident are the only 2 cards present.

That's when I found this thread over at VOGONS: https://www.vogons.org/viewtopic.php?t=48877
Interestingly enough, the OP over there was complaining about the exact same behavior, just without the extra complexity of another SCSI card. SO, in that thread the topic of driver release came up and was noted that older releases were more stable. I ended up downloading ASPI4DOS v3.35 (through the link in the thread) and....

IT WORKS! The driver loads, detects the 1542CF on IRQ 9, DMA 5, APSICD.SYS loads just fine, and the SCSI CD drive is picked up by MSCDEX.EXE/SHSUCDX.EXE to add a drive letter/fully accessible in DOS.

So, in summary:

• IRQs matter.
  • The 1542CF apparently just does not like IRQ 11 but works perfectly fine on IRQ 9
  • PCI based Ultra/Ultra-2 cards work fine on IRQ 11
• Driver release matters
  • I don't know what's wrong with the newest/currently available drivers from https://storage.microsemi.com/en-us/support/_eol/isa/aha-1542cf/ but they do not work very well
  • Use the APSI4DOS.SYS in the zip attached to this post instead

 

[NJRoadfan]

Weird that the 1540CF has problems with the latest ASPI driver. The 1540B is usually the troublesome one due to some rarely implemented hardware checks: https://www.os2museum.com/wp/learn-something-old-every-day-part-ix-aha-154xb-and-aspi4dos-sys/

Note that the aspi4dos.sys driver does have a verbose mode. Try using the "/D" switch and seeing what happens when the card is set to IRQ 11. Also, the 1540CF is NOT ISA Plug-and-Play compatible, you must reserve the IRQ you are using in the machine's BIOS. Under the PCI/PnP menu, set the IRQ you are using to "Legacy/ISA." This will prevent the BIOS from allocating the IRQ to other PCI or ISA PnP cards in the system at boot.

 

[Stinkerton18]

Weird that the 1540CF has problems with the latest ASPI driver. The 1540B is usually the troublesome one due to some rarely implemented hardware checks: https://www.os2museum.com/wp/learn-something-old-every-day-part-ix-aha-154xb-and-aspi4dos-sys/

Note that the aspi4dos.sys driver does have a verbose mode. Try using the "/D" switch and seeing what happens when the card is set to IRQ 11. Also, the 1540CF is NOT ISA Plug-and-Play compatible, you must reserve the IRQ you are using in the machine's BIOS. Under the PCI/PnP menu, set the IRQ you are using to "Legacy/ISA." This will prevent the BIOS from allocating the IRQ to other PCI or ISA PnP cards in the system at boot.

I wasn’t aware of the hardware checks but the P90 system I was using definitely does not have a Plug and Play BIOS. I have to set which ISA slots the PCI ones can share with.

 

[NJRoadfan]

A circa 1995 machine with a PCI bus should have a PnP BIOS. Aside from really early 1993 PCI boards that configured IRQs with jumpers, it should be PnP.