NeXTStep 3.3 and sparcstation 10

Started by mauve, January 12, 2009, 10:52:47 AM

Previous topic - Next topic

mauve

First, I wish to address my warmest thanks to all the contributors whose FAQs and documents were priceless in solving most of the problems I ran in when setting up my system ; I wouldn't have been able to do it without this help.

But let's reel back a little. I had that nice 10 / 512 MP with 179 MB ram sitting in my closet for more than a year, on which I used to run a webserver (debian linux). Having previously tried Solaris too, I thought I had quite a good grip on its quirks, and when I found some NeXTStep 3.3 CDs, I thought the time had finally come to revive it. I felt confident it would be nearly flawless. For some reasons, I've always been attracted to NeXTStep, ever since I read the 1988 account of the famed presentation of the system to the world in a computer mag, but never had the chance to run it first hand.

So I put my station back on line, made a few sanity checks with linux (still installed), and thought I was good to go.

details : the OBP is 2.14, disk 1 & 3 are 2 x 1GB for linux, disk 5 is a 2GB I was willing to dedicate to NeXTStep, and finally I had a cdrom unit as id #6.

1) my cdrom reader turned out to be unable to read the disks. Possibly shot, I had no other scsi unit with 512 Bytes/sectors capacity to try. So I scp'ed the iso to the station and dd'ed it on another disk unit, salvaged from a retired AS/400 RAID (the disk is 500 MB), and already re-sectored (AS/400 sectors are 520 bytes long). Hooked it as #6, reset, boot cdrom : no problems to select #5 as target for install. So far, so good.

2) First reboot. The doc I found in Apple KB says to boot on a disk different from #3, you can issue 'boot diskX' at the OBP prompt, where X is your unit #. Well, wrong, certainly not on my 10. I had to issue the whole device path, 'boot /iommu/sbus/espdma/esp/sd@5,0' ; took me a while to figure out, so I leave it here for posterity.
EDIT:
I had a chance to look at some Sun documentation about the OBP : only the first 3 scsi ids are bound to a diskX alias. Don't ask me why. Maybe - I'm reluctant to rebuild the system only to prove that point - this is the root (pun intended) of the subsequent mach failure to properly mount the newly installed disk as / (see further down).
/EDIT

3)Installation goes on flawlessly, time to reboot again.

4)I get a warning window with only "esc, intr, kill" and I'm thrown to a prompt. Why ? I type "exit", and the boot process grinds a bit before stalling. Hu Ho. Not good.

(some fiddling later)

77 :wink: ) I managed to pass the -v option to the kernel ; damn, mach is looking to mount the first disk of the scsi chain on '/', and discards the unit #5 on wich it lives ! Oh, but there is the rootdev option ! I'm saved !

108) NeXTStep acts as if it can't care less about the rootdev option ; strange. It insists on being on the first disk. Out of hope, let's try the -a (ask ?) option ! Now, the boot command is something like "boot /iommu/sbus/espdma/esp/sd@5,0 mach_kernel -v -a"

172) It doesn't cut it either. For some reasons I can't explain, at the exact moment mach requests me to enter the rootdev, my keyboard ceases to function. Not once or twice, every single time ! I can't type anything.

So I've removed the 2 internal disks, only leaving the #5 attached. And finally, I was able to discover NeXTSTEP 3.3 in all its glory.

Question : did I do something wrong, or did I forgot something important ? The situation is not satisfactory, with the demise of my cd reader. I wish I could switch between linux and NeXTSTEP without a screwdriver.

EDIT :
I finally nailed it : I edited /etc/fstab and set sd2a as root. I then put back my internal disks where they belong, and started as usual. It works fine except for the warning as the disks are not in a known format. I've seen elsewhere the solution to this, so it's a mere annoyance that will be (hopefully) shortlived. All is well that ends well ? Not really. In the process of digging into the system, I discovered that the install process set up the disk as a 1GB unit instead of the 2GB available. So at the moment, I've 1GB missing, and don't know where the h*ll this space has gone...
/EDIT

But anyway, for what I saw, NeXTSTEP is awesome, and I'm truly happy to have managed the process so far. I write this experience also because I have a feeling the 10 being older than the 5 & 20, it has its own quirks, and I may very well have hit long forgotten bugs of that specific station.

EDIT :
After reading somewhere that a SuperSparc II SM71 would greatly boost NeXTstep on my hardware, I hunted one and got it from a really kind fellow nextstepper who was willing to part with one of his. At first, it was of no use because the SM71 really needs a prom > 2.22 to even *boot* the station. But said person was kind enough to burn a 2.25 prom for me, and I just swapped my 2.14 in favour of the newer one and, while it doesn't really makes the system fly (not expected with 75 MHz cpu anyway), it sure gives a bit of a kick. Overall, a recommended upgrade. I didn't took the pain to reinstall the system, it works as is, but does anybody knows if making a fresh install is recommended ? I was guessing perhaps a more optimised kernel image could be installed when a supersparcII was detected in the process (just a wild guess, haven't read anything on the subject) ?
/EDIT
Thanks again, your posts more than helped me.