Hi all,

In my world (Ubuntu, Windoze, Amiga(?)) you can put in a new disk drive in a computer, and remove it without affecting the computer. Of course if it's not the boot drive.

Now. After i succesfully configured that 4G (partitioned) Sandisk CF card on the A4000 (not on WinUAE), the computer will not boot if either the cf card or the whole adapter is not present.

WTF???

The computer in question:

A4000 desktop, KS and WB 3.0
WarpEngine 040, with 1 drive on SCSI bus
Mediator with Radeon, SpiderII, NetCard
Original IDE drive

I saw the word "sandisk" in the hexdump of the system file "drivedefinitions" but did not try to hide the file. i did not try to boot from (the external, as the original internal is long gone) floppy either.

Any ideas?

ThanxThinkTank,

Claude.

@mossieurclo

Check the IDE drive's jumpers. It might have different settings for master without slave (single) and master with slave present.

Another culprit could be Kick 3.0. It is known to overlook an IDE harddrive at the first try. Press Ctrl-A-A. If if boots then, an upgrade to Kick 3.1 might help.

The drive definitions file is not needed for anything. It is just a history of all harddrives you ever initialised with HDToolbox. You can safely delete it.

_________________
Email: thomas-rapp@web.de
Home: thomas-rapp.homepage.t-online.de

@mossieurclo

I have seen this sometimes when "installing" a new disk in the chain or whatever it is called. Then when it is removed, the original disks cannot be found because of a missing item in the chain. Might be the same issue thomas was talking about.

In this case one solution would be to not save/update the existing disks even if you are asked to do so by HDToolbox when preparing the new disk/card.

_________________
- Don't get fooled by my avatar, I'm not like that (anymore, mostly... maybe only sometimes)
> Amiga Classic and OS4 developer for OnyxSoft.

@Deniil715

Quote:

In this case one solution would be to not save/update the existing disks even if you are asked to do so by HDToolbox when preparing the new disk/card.

That answer spooked me out! Because Clémentine (that's the A4000D we're discussing. Clémentine, meet thomas and Deniil, helpful people on this forum ) would not even SEE the CF, until i resolved to heed HDToolbox's warning: "(your original ide HDrive) has been modified. New drives could go undetected". The original ide had NOT been modified, and i was afraid i'd loose data if i would "save changes to drive".

But at one point i noticed the "new drives could go undetected" part, was still afraid to loose data, but hit the "save changes" button anyway. Expecting a warning "data will be lost", but none appeared. Next boot was normal to my immense relief, and i could then see and configure the CFCard. Then were raised the questions in my original post.

@thomas

Quote:

Check the IDE drive's jumpers. It might have different settings for master without slave (single) and master with slave present.

So you started me playing with jumpers, and that did it.

The weird thing is that on the wdac2200 (Caviar2200, yes, the drive that came with this computer in 1994!!) the J8 jumpers are labeled ma, sl, cp. cp??? i read documents that said "do not shunt 1-2 (cp) it is reserved", other documents indicated it was the CableSelect setting.

So i tried it (set the Caviar AND CF adapter to CS) and it worked. I could remove the CF from its slot on the adapter, or remove the whole adapter, and Clémentine would boot.

Strangely though, the boot after CFCard removal takes

Last edited by mossieurclo on 30-Jan-2015 at 11:01 PM.

(post trucated??? the rest follows. Am i that too much verbose?)

more than 60 secs. Also i thought that the A4000 IDE stuff (SCSI.device?) was not modern enough to support CableSelect.

But hey, it worked!

(
After noticing the long boot time after CF removal, i pushed the experiment, thinking that maybe i did not wait long enough for boot on the first trys before playing with jumpers. So i reset the drives as master and slave as it was, and it would not boot after CF removal. Reverting to CS setting worked again (with the long boot event) afterward.
)

Thanx for your input everyone, sorry for the (seeding Google for someone else's search) long reply.

Claude

Last edited by mossieurclo on 30-Jan-2015 at 11:07 PM.

@mossieurclo

You're post wasn't that big. Should have been able to add on bottom.

I don't know what controls CS but I would have thought it was in hardware. Until your story I hadn't seen reason to be afraid of saving changes to drive.

But it is obvious that if you have a master/slave setup removing one will break it. Except a master by itself should be okay.

@Hypex

Quote:

You're post wasn't that big. Should have been able to add on bottom.

I tried editing in adding the rest to the bottom but it did not work. Probably the "greater than" chevron before "60 secs" that blocked the rest (i imagine the bulletin board system expecting a "less than" chevron).

Quote:

[...] Until your story I hadn't seen reason to be afraid of saving changes to drive.

There IS no reason. i've seen that there's no reason to be afraid of "save changes to drive". It does not wipe any data, contrary to what i was afraid of. The weird thing is that that drive had NOT been modified, so i tended to ignore the warning. Hmm.

Quote:

[...] Except a master by itself should be okay.

Exactly the point of my original post. If i remove the second (slave) drive, the one that has always been there (the original AC2200 jumpered as master) should suffice, since it was the original set up of the machine. Weird.

Thanx for your thoughts. Dodo time (i had a to run the the snow mower today, haan).

Claude.

@mossieurclo

Perhaps it got confused bv the chevron and thouight you were trying to open up a stargate.

But there is one case to be afraid of when saving changes to drive. And that would be to initialise it by reading in the drive definitions and resetting the partition layout. Of course that would be a bit drastic and you would know to be careful there.

I wonder if it gets confused by the last drive set. Then again if CS mode works fine for "hot-switching" the drives it should be fine. Well at least it works now.

@Hypex

Quote:

@mossieurclo Perhaps it got confused bv the chevron and thouight you were trying to open up a stargate.

Beware, i CAN be that clumsy. And the whole adventure of rejuvenating Clémentine (details in another post) sometimes feels like i am indeed trying to open up a stargate

Quote:

But there is one case to be afraid of when saving changes to drive. And that would be to initialise it by reading in the drive definitions and resetting the partition layout. Of course that would be a bit drastic and you would know to be careful there.

Yes, i should have insisted that "save changes to drive" will not loose data IF you did not change anything on the drive before hitting that button. Which was the case and had me wonder why HDToolbox saw the drive as "modified"

Quote:

I wonder if it gets confused by the last drive set.

I am also wondering about that long delay in "redetecting" the drives chain.
To be precise, it has that delay on the next boot after i remove the CF card. The delay happens BEFORE the Early Startup gray screen apppears. If i put the CF back in, boot is normal.

So for the disk detection routine to be disturbed, the previous state has to have been written somewhere, right? My question is where?? At that stage only the ROM and first few blocks of a disk are "live". Where is the info about the 'supposed' state of the drives chain written??
The only hints i have is the few accesses to the Caviar i hear during that delay.

Quote:

Then again if CS mode works fine for "hot-switching" the drives it should be fine.

Speaking of which: can we "hot swap" CF cards on Amigas? I mean removing a card without powering down. The adapter is a Startech 35BAYCF2IDE if it matters.

Quote:

Well at least it works now.

Well at least it works now.

ThanxThinkTank,

Claude

@mossieurclo

I've accidentally unplugged a harddrive (actually, the only one, ie the one with the sys partition on it) on my A1200 back in the day. Worked like a charm. There's a bit that says if the drive has removable media, but Amiga doesn't seem to respect this, and just treats everything as removable. As far as I could tell, anyway.

Worst case there's the "diskchange" cli program if the new one isn't detected.

On the other hand, this is A SERIOUSLY STUPID IDEA, AND SHOULD NOT BE DONE BY ANYONE

(Edit: This posting is akin to giving people a loaded shotgun and a map to where there foot is, but with the writing "don't shoot this part" in small letters).

Last edited by olegil on 02-Feb-2015 at 12:13 PM.

_________________
This weeks pet peeve:
Using "voltage" instead of "potential", which leads to inventing new words like "amperage" instead of "current" (I, measured in A) or possible "charge" (amperehours, Ah or Coulomb, C). Sometimes I don't even know what people mean.

@olegil

Quote:

On the other hand, this is A SERIOUSLY STUPID IDEA, AND SHOULD NOT BE DONE BY ANYONE

Erm... WHAT is a stupid idea?

@mossieurclo

Good you managed to fix it. When saving this "modified disk" it updates the Last LUN flag, I think. The Last LUN tells the IDE scanner when to stop, or if it should continue to try a longer time on slots where there should be a disk.

@olegil

My experience on hotswapping IDE is not good. I've seen lockups and wierd behaviour when I have more or less accidently removed an IDE disk. Might work if you make sure not to touch any partition while it's gone, but scsi.device does not support a "disk not present" flag AFAIK.

_________________
- Don't get fooled by my avatar, I'm not like that (anymore, mostly... maybe only sometimes)
> Amiga Classic and OS4 developer for OnyxSoft.

@Deniil715

Quote:

Deniil715 wrote: @mossieurclo Good you managed to fix it. When saving this "modified disk" it updates the Last LUN flag, I think. The Last LUN tells the IDE scanner when to stop, or if it should continue to try a longer time on slots where there should be a disk.

And that Last Lun flag is saved on the RDB, i presume? And if i experiment with the values of that flag to prevent that long delay, i will have to "save changes to disk". Will the data stay? i would guess yes as long as i dont' change the partitions, but any other advice?

StillThanxThinkTank

Claude

@mossieurclo

What time is it? Clémentine.

Quote:

Which was the case and had me wonder why HDToolbox saw the drive as "modified"

It probably scanned the busses and found a new HD. Most likely if the main HD is set at last HD and it finds another one.

Quote:

I am also wondering about that long delay in "redetecting" the drives chain.

How long is the delay? There are usually two reasons for a delay. First is for no devices or HD found on the internal IDE and it can look for 30 seconds. The second is a slow HD spin up time but I think we can skip that one, if your drive isn't that old.

Quote:

To be precise, it has that delay on the next boot after i remove the CF card.

Do you mean with the power still on? If so that would stuff the system up!

Quote:

My question is where??

Those settings would be stored in the RDB.

Quote:

Speaking of which: can we "hot swap" CF cards on Amigas?

Only on the side! Like an A1200 CF slot. For a CF hooked up on the internal IDE I would say no. IIRC the IDE is unbuffered so a bit unsafe on the electronics. Apart from that you are pulling a drive out of the system which is not removable. Yes it's easily removable for you but the system sees it as a HD that is permanent.

That said I have accidently pulled SCSI drives off a chain and the Amiga did quite well dealing with that as well as IDE when I also did so accidently. Doing it on a PC didn't do so well!

There are exceptions. For example, if you decide to install the OS3.9 FFS to RDB to a HD, only to find it crashes on bootup you will need to remove by force. And that is by booting up from floppy without the HD, then plugging it live into the IDE port, so you can then run HDToolbox and wipe it off.

Last edited by Hypex on 07-Feb-2015 at 12:59 PM.

@mossieurclo

Quote:

mossieurclo wrote: And that Last Lun flag is saved on the RDB, i presume? And if i experiment with the values of that flag to prevent that long delay, i will have to "save changes to disk". Will the data stay? i would guess yes as long as i dont' change the partitions, but any other advice?

Here is a program which lets you examine or change the RDB flags: without HDToolbox: http://thomas-rapp.homepage.t-online.de/downloads/changerdbflags.lha
A similar program for partition flags exists, too: http://thomas-rapp.homepage.t-online.de/downloads/changebootpri.lha

_________________
Email: thomas-rapp@web.de
Home: thomas-rapp.homepage.t-online.de

@Hypex

Quote:

Hypex wrote: There are exceptions. For example, if you decide to install the OS3.9 RDB to a HD, only to find it crashes on bootup you will need to remove by force. And that is by booting up from floppy without the HD, then plugging it live into the IDE port, so you can then run HDToolbox and wipe it off.

This is dangerous nonsense. Firstly there is no such think like an OS 3.9 RDB. The RDB format has not changed since it was invented.
And secondly HDToolbox won't recognise such a late-plugged-in drive without a reboot. It would probably just hang because the bus was crashed by connecting electronics with power already on.

HDToolbox does not "scan the bus" or whatever you imagine. It just tries to open scsi.device units one after the other. And scsi.device only opens on those units which were found at boot time, not later.

The only occasion where I got a hot-plugged drive to be recognised is with those SCSI controllers made by Phase5 which are supported by the UnitControl utility. The utility sends a secret command to the SCSI driver which causes it to rescan the bus. But this command is internal to Phase5 and does not work with drivers made by other manufacturers.

_________________
Email: thomas-rapp@web.de
Home: thomas-rapp.homepage.t-online.de

@Hypex

Quote:

For example, if you decide to install the OS3.9 RDB to a HD, only to find it crashes on bootup you will need to remove by force. And that is by booting up from floppy without the HD, then plugging it live into the IDE port, so you can then run HDToolbox and wipe it off.

I thought you could boot from the floppy without removing the HD????from early startup at least. And then bring up HDToolbox and fix the drive?

Anyway thanx for your input everyone.

I'll be away for 21 days, driving that semi truck around.

Claude

@thomas

Quote:

Firstly there is no such think like an OS 3.9 RDB. The RDB format has not changed since it was invented.

Sorry I meant the OS3.9 FFS. I've made the mistake of installing this into the RDB and it renders the Amiga useless on power on.

Quote:

And secondly HDToolbox won't recognise such a late-plugged-in drive without a reboot. It would probably just hang because the bus was crashed by connecting electronics with power already on.

I don't recall exactly how but I needed to plug in the HD after it had powered on and at a point where it could detect the HD but not load in the FFS off the RDB.

I know I didn't plug the HD into an A1 externally to wipe the RDB or similar but did it on the machine that crashed on bootup. IIRC I've seen the problem on both the A1200 and A4000. Very annoying when you think you are doing the right thing by installing a modern filesystem to the RDB only to be greeted with a circular crash on power up that prevents the machine working.

@mossieurclo

Quote:

I thought you could boot from the floppy without removing the HD????from early startup at least. And then bring up HDToolbox and fix the drive?

You can if it still powers on and boots up fine. But in my case I had installed the OS3.9 FFS to RDB which caused an instant crash at power on. It just kept crashing as soon as it read the HD so could not boot at all. Reset causing the same thing.

(i'm baaaa,aaack. Despite the last 2 days running with the red telltale announcing "this engine will be shut down soon". Stressfull at -20 celcius. )

@Hypex
Quote:

[...] But in my case I had installed the OS3.9 FFS to RDB which caused an instant crash at power on. It just kept crashing as soon as it read the HD so could not boot at all. Reset causing the same thing.

Did you learn what was wrong with that FFS? Were you running KS3.1?

I was thinking of experimenting with other file sytems (PFS or whateverelse), and you make me think twice...