One of the problems I have had with a SPARC-clone that I bought on Ebay was that the system was password-protected : firmware commands could not be executed without entering a password. Unfortunately, the seller was unresponsive.
One way of fixing a password problem on a SPARC is the following:
- Power up machine
- Press 'STOP-A' to go into firmware command mode
- Enter a command to be prompted for a password
- Press <ENTER>
- While the firmware is waiting for 10 seconds because of the bad password, yank out the NVRAM device
- When the command prompt returns, enter a command. This time, the command will not demand a password.
- Re-insert the NVRAM device.
- Do "setenv set-defaults true"
The problem with yanking out the NVRAM device is that you might destroy it or your motherboard or both. Since a new or used NVRAM can cost $20 or more, you might want to think twice.
The solution is to built a little contraption that will remove power to the NVRAM without the need to physically remove it i.e. use a switch!
See the picture below for my solution. What I have are three cheap 28 pin IC sockets (my SPARC clone used a 48T08). Pin 28 of the 48T08 is the power supply (Vcc, in electronics-speak). Pin 28 of the top socket is bent out and a wire is soldered to it.. Pin 28 of the middle socket is removed : this acts like an isolation layer. The bottom socket has a wire soldered to the socket. Both wires are soldered to a slide switch.
The 48T08 is plugged into the top socket and the entire contraption is plugged onto the motherboard.
Now, to remove the password, the above method is used. The main difference is that instead of physically removing the NVRAM device, you switch it off!
I had to connect to a serial console in order to boot the machine that had the NVRAM problem. This was simpler than I thought.
I obtained a commercial null modem cable with 9-pin at one end, and 25-pin at the other end. I needed a gender convertor to convert the 25-pin end of the cable to a male so that it would fit the serial port A of the SPARC.
With that configuration, I started up Hyperterminal on Windows 98. I set the serial port to 9600 baud/bits per second, 8 bits, no parity, 1 stop bit and with hardware handshaking.
When I powered up the SPARC, I had to wait for a few seconds before the initial sign on messages appeared.
STOP-A is pressing the L1 and A keys simultaneously. On older keyboards, the STOP key is marked as L1.
How do you do A-L1 or STOP-A with a serial console in order to stop the boot process and enter the ROM monitor?
With Hyperterminal under Windows 98, pressing CTRL-BREAK will do the trick.
For reasons best known to themselves (Note 1), the folks at SUN decided that the default boot SCSI hard disk must be drive 3 : if you want to boot from a SCSI hard disk, you have to configure it (via jumpers) to have a SCSI ID of 3.
This creates problems when you use third party operating systems like OpenBSD which detect and number drives from ID 0 onwards.
The boot drive has an ID of 3. If you add a drive with an ID of 0, OpenBSD will see ID 0 as the first drive and ID 3 as the second drive and will fail to boot. Therefore, the second drive needs to have an ID of 4. A third drive needs to have an ID of 5. You do not want to use ID 6 because that is for the bootable SCSI CDROM drive.
Does that mean that you are limited to 3 SCSI hard disks under OpenBSD?
No. What you need to do is to be able to configure the SPARC Station so that it will boot from ID 0. That will allow you to add drives with ID 1, ID 2 and so on until ID 5 (with ID 6 still available for a CDROM drive).
This is done via the firmware with the "set boot-device" command. I am assuming a SPARC LX or better machine.
First, type "boot sd(0,0)"
This will result in an error message with a suggestion to :
Type "boot xx() " to boot from the defaults:
boot sd() = /iommu/sbus/espdma@4,8400000/esp@4,8800000/sd@3,0
Note down everything after the '=' sign. It is different for every type of machine.
Now type in :
setenv boot-device /iommu/sbus/espdma@4,8400000/esp@4,8800000/sd@0,0
The /iommu ... string is what you noted down. The only change from the default is that you now use sd@0,0 instead of sd@3,0
This will tell the SPARC to boot from ID 0 instead of ID 3.
An Axil 235 machine is a semi-clone of the SPARC 10. It uses the same chipset and has a single MBUS slot. This means that theoretically, you can swap the standard 40MHz CPU for a faster/better one.
In practice, the original BIOS (Note 2) does not support the new SuperSPARC or HyperSPARC CPU cards.
Finally, I went out and got the Version 2.25 BIOS upgrade for the SPARC 10. My reasoning was : the Axil is very much like the SPARC 10 with the same motherboard chipset, the 80MHz and 72MHz crystals, the 8 memory sockets. The only differences are that the Axil only has one MBUS slot and there is an onboard cgsix video.
I plugged in the new BIOS ROM, plugged in the SM71, powered on the system and ... nothing happened.
I removed the keyboard and connected a serial terminal, powered on and the diagnostic messages came on! The system completed power-on self tests and tried to boot from the network. It appeared that everything was working.
I plugged in the keyboard and powered on the system again. Diagnostic messages appeared on the terminal. But there was NO VIDEO!
Then I realized something : the Axil had onboard video. The video BIOS was part of the original Axil BIOS ROM. When I swapped in the SPARC 10 BIOS ROM, there was no more video BIOS. No wonder the system booted without video!
I plugged in a spare cgsix card. And voila! I had video!
To test the system further, I removed the SM71 and plugged in a 125MHz ROSS HyperSPARC. The system booted!
Conclusion : The Axil 235 can use faster and more powerful CPUs. However, you have to use the 2.25 SPARC 10 BIOS and an SBUS video card.
The SPARC 10 BIOS version 2.10 does not recognize the SM71 SuperSPARC CPU. You have to upgrade to the Version 2.25 BIOS. I got mine from www.memoryx.com.
In addition to the Axil 235, I have two other SPARC clones. They were both made by Super Workstation Inc, a now defunct manufacturer of clones. Some say that it was a Korean or Taiwanese company.
The first machine is a SuperWorkstation LS. This is equivalent to a SPARCStation LX but in a SPARCStation 2-type pizza box.
The large chip on the lower right of the last picture is the CPU. It is a TI MicroSPARC running at 50MHz. The chipset that is used is from NCR. Note that it is exactly the same chipset used in the Improv5 motherboard that this website was running on.
For the uninitiated, the Improv5 motherboard is an upgrade for a SPARCStation 1,1+,2,2+ and contains the same CPU that is found in the SPARC 4 & 5 machines. Essentially, you get a machine that is functionally equivalent to a SPARC 4 or 5. In addition, 32MB 72-pin SIMM memory modules can be used for a total of 192MB onboard memory.
On this SuperWorkstation LS, it might be possible to upgrade the firmware and swap in a SPARC 5 CPU. But I have not tried that out.
The other machine is a SuperWorkstation II+/50. Click here for the picture. It is a pizza-box unit. The one I have does not boot. It looks like a SPARCStation 2-class machine, perhaps running at 50MHz. It accepts 30-pin SIMMs. Picture of the inside.
Recently the company I worked for dumped a SPARC 20. No, I didn't managed to get my hands on it, but a coworker did and he promptly wanted to put Debian Linux on it.
We know that the old SPARCs need a special CDROM drive in order to boot from a CDROM.
The question then was can a CDROM that is burned on a PC be used to boot a SPARC (assuming it contained a bootable image)?
My coworker downloaded the ISO image of Debian Linux for SPARC and used his PC (running Linux) to burn a CDROM. The end result was that it worked!