Seeing same drive multiple times - how to fix?

[NorthWay]

After much wrangling with my CS060 I have found a way to see both my old drive and my new SSD via an Acard ARS-2000SUP adapter.

SCSIConfig was much confused. I gave it the numbers from HDToolBox instead.
I made a 250M "SS0:" partition. After a reboot I now have 8 new partitions: SS0: to SS7:. All with the same name ("Flash") now. This makes the OS much unhappy - booting it you can forget about.

Any hints on how I can make the ghost partitions go away?

(At first it was only SCSIConfig that saw multiple LUNs, but now HDToolBox does too.)

[idrougge]

You somehow need to set max LUN.

[NorthWay]

Something is stirring in the back of my head about that.
Is there a LASTLUN setting in the RDB that you can set? What tool would I use to do such a thing? (I think I have done this once before. Possibly.)

EDIT: I found http://aminet.net/package/disk/misc/RDBFlags-1.3
The question is if I have that one my hd already... (floppy doesn't work, CD-ROM might if I start swapping cables again, network card isn't in, IDE CD I have never used before but it gave me odd behaviour when connected. So getting stuff in and out is an uphill battle. Both ways.)

[nogginthenog]

I have the same problem with my ACARD AEC-7726Q.
Using a CyberStorm Mk II with SCSI module and SCSI wide adapter (I tried 3).

I had to disable reselection in the RDB (I guess you did that?).

See here:
http://eab.abime.net/showpost.php?p=...8&postcount=78

Never found a good solution

[NorthWay]

So I tried setting LastLUN. Nothing changed.
I tried setting Reselection. Machine kept booting. Possibly because of an 8000000B.
Didn't have time to scratch everything and try from start again.

Only thing I saw from the old 18G was that it had Reselection and Synchronous set. I'm pretty sure I did the same thing to it 10-15 years ago as SCSIConfig see all the 8 LUNs of it, but if I used RDBFlags I'm not sure.
I see I have some old stuff commented out in my s-s, among which is UnitControl to set NoFWC (I'll have to google that to refresh my memory).

[NorthWay]

Summary so far: Reselection on both drives gives eternal reboot loop without any guru.
Setting the "LastXXX" flags changes nothing.
My old 18G behaves like it is read-only. I write something to it, and after I boot it isn't there any more. I have tried using "NoFWC" with UnitControl without any more luck. I copy something to an SFS partition and it looks like random noise data when I look at the copy.

Next step is to see if I have some old memory tester and run that.
Plus I'll throw together some old PC parts that can grok one of the SCSI cards I have in my drawers and give the whole SSD drive to WinUAE and see if that can produce something edible.

[nogginthenog]

I think the multi LUN thing it's a limitation of cybscsi.device.
My ACARD 7726 works perfectly in an old XP machine with a SCSI card.

I also tried flashing different versions of the Acard software to the card.

[NorthWay]

Memory tested good. Still not done the WinUAE thing.

One tiny victory though: Setting the Master jumper on the IDE DVD-Writer now makes the machine boot instantly and not continually light up the drive "H.DISK" led. (I think I found the original IDE cable from my 4000 - I had gifted it to my C-One ages ago when I though I would use a cd-rom in it).

[NorthWay]

Tested every which way until the cows come home. Set up the SATA disk itself with WinUAE with config similar to my own machine; worked perfectly.
Put it back in the SCSI adapter and hooked up to 4000 - blinking power light without even getting a meditation before it cycles and blinks again. Turn off adapter power, boot 4000, power on adapter, "Unitcontrol rescan", and both HDToolBox/SCSIConfig see the partitions and stored filesystems as if everything was normal. The only thing I haven't tried is making mountlists to get the filesystems themselves mounted.

Anyone feel like ReSource'ing, debugging and fixing cybscsi.device? (Could bigger size math trigger some kind of 060-non-implemented bug?)

UPDATE: The 3000 booted straight up from it, apart from only showing 3 partitions until SetPatch had run.

[nogginthenog]

Originally Posted by NorthWay

Tested every which way until the cows come home.

The only thing I haven't tried is making mountlists to get the filesystems themselves mounted.

It works, but you need to boot from a different device. Mark the cybscsi partitions as non mountable, manually mount and modify the startup-sequence to continue from a cybscsi partition.

Anyone feel like ReSource'ing, debugging and fixing cybscsi.device? (Could bigger size math trigger some kind of 060-non-implemented bug?)

Probably all that is necessary is to disable SCSI LUN scanning.

It would be interesting to see if 68k/Linux works. It supports the CyberSCSI module. Might give it a try. I did have Linux running many years ago.

[NorthWay]

I turned off all automounts. I thought. Now we're back in Weird Country again.

When I boot up again the bootable and automount settings are back on again as they were. I doublechecked by starting HDToolBox a second time to see they were gone before the boot. This is the same weird thing I have with my old 18G drive. I write things to it, and next boot they have escaped through a black hole. Banging your head against a wall doesn't even begin to describe it.
LastLUN/LastDisk/LastUnit had no effect. That one seemed to survive from boot to boot last time I fiddled with it.

Back to Plan B: Use the IDE disk (only have boot partition so far). Fire up ReSource when done.

[Toni Wilen]

Does early boot menu open and can you boot without startup-sequence? (Because reboot loop and/or 8000000B guru is "normal" when using 68060 with 68040 CPU libraries)

About LUNs, I think later Amiga SCSI controllers use SCSI identity messages to send LUN, most older style SCSI devices put LUN in SCSI command block. It is possible your adapter ignores SCSI messages. (WinUAE supports both and always rejects non-zero LUNs, it is not configurable)

[NorthWay]

Originally Posted by Toni Wilen

Does early boot menu open and can you boot without startup-sequence?

No. No. The blinking guru reboot cycle will only break when I power off the disk enclosure after which it finally gives 8000000B guru with cybscsi.device task address as the guru address.

Hooking the SSD up through SATA in my PC I emulate a 4000+CS060+CybSCSI and I gave WinUAE the whole drive. The emulated cybscsi.device was happy enough with that, but it gave me some other head/cylinder etc settings than what HDToolBox natively does (SCSIConfig insisted on 0 heads and similar nonsense in the 4000).
Put the disk back in the Acard and in the 3000, and it boots straight up.

I have now pulled the SCSI module to make more room for cabling. My 4000 has a working setup with an 8G IDE disk. Time to insert cards and see if I can get online and install drivers and updates.

[NorthWay]

UPDATE:
cybscsi.device 8.7 does not fall over.
Still not found a way to hook up internally, but the external is ok.
I still get 7(8?) partitions for every real partition (i.e. AGA: also becomes AGB: AGC: etc, and they all have the same name so 7 partitions named "Main:" - which confuses the hell out of the OS). These can be disabled in the boot menu (which is too annoying other than for testing when there are 9 real partitions, i.e. 9*7...).
I patched away a loop in the device driver which I think was the LUN handling, and now it boots as normal as you can get it (ok, my first attempt at using the boot menu made it fall over with a mega-requester throwing stats at me, but I can probably live without it).

[kolla]

Looks almost like you have accomplished multipath on Amiga, congrats, lol