Very Computer

A wierd problem here ...

486/66dx2, 32 megs RAM, UMC (?) chipset, AMI Bios.
It's one of those funky MB's with 4 30 pin memory slots and 2 72 pin
slots, currently with two 16 MB 72 pin SIMMS.

I can't find my documentation on it, unfortunately.  I can't find
anything to help me figure out the chipset on the MB, though at least
on chip says UMC.  It's a VL-Bus MB, though it doesn't have any VLB
cards in it.

When I put in the Adaptec 1542B, they keyboard doesn't work at all -
not even the numlock or capslock lights work.  I can't even hit escape
to speed up (skip) the memory test.  I've even taken out all the cards
except for the 1542B and the video card (and I've tried a few
different video cards) and the same thing happens.

If I disable the "quit on keyboard error" option, the machine will
boot up with the keyboard not working, and Linux will even use the
1542B - I can log in remotely (via the network) and everything seems
ok except for the keyboard.

If I remove the 1542B, the keyboard works fine.

I've tried numerous combinations on the IRQ, Base address, etc. to no
avail.  I don't seem to have the parameters to change the BIOS address
(I don't have the documentation on the 1542 either, just a Microhouse
printout, and I might be missing a page on that)

I've also tried an 1542CF, and it has at least once booted up with the
keyboard not working as well, though a powercycle seemed to fix that.
Unfortunately, Linux does not want to work with that card for some
reason, no matter *what* I do.  I've not tried DOS, as I really have
no plans to use the SCSI disks under DOS (the machine also has IDE
drives).

Anybody have any idea what's going on here?  This thing is driving me
nuts!  Thanks in advance ...

--
For further information, please e-mail the Department Of Useful
Guidelines on Modern Communications (DOUGMC).  Their e-mail address

"Kirk to Enterprise -- beam down yeoman Rand and a six-pack."

Maybe some problem related to the fact that 1542 does bus mastering on an ISA
bus and can't address al memory flat, and you have 32 MB. Try a local bus
controller ...  Stefan

: A wierd problem here ...

: 486/66dx2, 32 megs RAM, UMC (?) chipset, AMI Bios.
: It's one of those funky MB's with 4 30 pin memory slots and 2 72 pin
: slots, currently with two 16 MB 72 pin SIMMS.

: I can't find my documentation on it, unfortunately.  I can't find
: anything to help me figure out the chipset on the MB, though at least
: on chip says UMC.  It's a VL-Bus MB, though it doesn't have any VLB
: cards in it.

: When I put in the Adaptec 1542B, they keyboard doesn't work at all -
: not even the numlock or capslock lights work.  I can't even hit escape
: to speed up (skip) the memory test.  I've even taken out all the cards
: except for the 1542B and the video card (and I've tried a few
: different video cards) and the same thing happens.

: If I disable the "quit on keyboard error" option, the machine will
: boot up with the keyboard not working, and Linux will even use the
: 1542B - I can log in remotely (via the network) and everything seems
: ok except for the keyboard.

: If I remove the 1542B, the keyboard works fine.

: I've tried numerous combinations on the IRQ, Base address, etc. to no
: avail.  I don't seem to have the parameters to change the BIOS address
: (I don't have the documentation on the 1542 either, just a Microhouse
: printout, and I might be missing a page on that)

: I've also tried an 1542CF, and it has at least once booted up with the
: keyboard not working as well, though a powercycle seemed to fix that.
: Unfortunately, Linux does not want to work with that card for some
: reason, no matter *what* I do.  I've not tried DOS, as I really have
: no plans to use the SCSI disks under DOS (the machine also has IDE
: drives).

: Anybody have any idea what's going on here?  This thing is driving me
: nuts!  Thanks in advance ...

: --
: For further information, please e-mail the Department Of Useful
: Guidelines on Modern Communications (DOUGMC).  Their e-mail address

: "Kirk to Enterprise -- beam down yeoman Rand and a six-pack."

|> A wierd problem here ...
|>
|> 486/66dx2, 32 megs RAM, UMC (?) chipset, AMI Bios.
|> It's one of those funky MB's with 4 30 pin memory slots and 2 72 pin
|> slots, currently with two 16 MB 72 pin SIMMS.
|>
|> I can't find my documentation on it, unfortunately.  I can't find
|> anything to help me figure out the chipset on the MB, though at least
|> on chip says UMC.  It's a VL-Bus MB, though it doesn't have any VLB
|> cards in it.
|>
|> When I put in the Adaptec 1542B, they keyboard doesn't work at all -
|> not even the numlock or capslock lights work.  I can't even hit escape
|> to speed up (skip) the memory test.  I've even taken out all the cards
|> except for the 1542B and the video card (and I've tried a few
|> different video cards) and the same thing happens.

Try turning off the "Fast A20 Gate Operation" item in the AMI BIOS
advanced CMOS setup. One of the many spectacular design innovations
in the original IBM PC-AT was routing one of the memory address lines
(the A20 line) through the keyboard controller chip. This has been
known thoroughly confuse some bus mastering cards.

|> If I disable the "quit on keyboard error" option, the machine will
|> boot up with the keyboard not working, and Linux will even use the
|> 1542B - I can log in remotely (via the network) and everything seems
|> ok except for the keyboard.
|>
|> If I remove the 1542B, the keyboard works fine.
|>
|> I've tried numerous combinations on the IRQ, Base address, etc. to no
|> avail.  I don't seem to have the parameters to change the BIOS address
|> (I don't have the documentation on the 1542 either, just a Microhouse
|> printout, and I might be missing a page on that)
|>
|> I've also tried an 1542CF, and it has at least once booted up with the
|> keyboard not working as well, though a powercycle seemed to fix that.
|> Unfortunately, Linux does not want to work with that card for some
|> reason, no matter *what* I do.  I've not tried DOS, as I really have
|> no plans to use the SCSI disks under DOS (the machine also has IDE
|> drives).
|>
|> Anybody have any idea what's going on here?  This thing is driving me
|> nuts!  Thanks in advance ...
|>
|> --
|> For further information, please e-mail the Department Of Useful
|> Guidelines on Modern Communications (DOUGMC).  Their e-mail address

|> "Kirk to Enterprise -- beam down yeoman Rand and a six-pack."

--
  Joseph S. Wisniewski  |  The views expressed are purely my own, and do not
   Ford Motor Company   |  reflect those of the Ford Motor Company, or any of
    Project Sapphire    |  its affiliates.

Quote:>Maybe some problem related to the fact that 1542 does bus mastering on an ISA
>bus and can't address al memory flat, and you have 32 MB. Try a local bus
>controller ...  Stefan

That isn't the reason.  Linux can work around this using double buffering.

Rob