workbench3.1 and SFS - incorrect partition sizes

[rvo_nl]

Hi there, its me again!

Made a lot of progress since I started a month ago:

- Found the correct way of installing OS3.9, both Boing Bags, CyberGraphX, PowerUP, WarpOS and Warp3D
- Installed internet and filesharing
- Enjoyed OS3.9 for a while, then decided it was time to repartition and install SFS. I've done this, and the following is the way my drive is now partitioned (using HDToolbox of OS3.9):

HD0/System: 100MB, FFS Workbench 3.1
MAC: 2.5GB, FFS
PC: 1.4GB, MSH\00
------------------------------------4gb barrier
SWAP: 1GB (I wont be using this, its just to be safe)
HD1/OS3.9: 10GB, SFS2, AmigaOS3.9
HD2/OS4: 15GB, SFS2, AmigaOS4.0
HD3/Games: 50GB, SFS2, Games
HD4/Data: 30GB, SFS2, Data
HD5/Spare: ??GB, SFS2, Backup

WB3.1 will be my boot partition. I will write a little script that allows me to select one of the 3 OS's to boot from. But for this to work, all partitions must be correctly mounted under Workbench3.1. And that's where I'm stuck now.

- I have an IDE-Fix express, so c:IDEfix is loaded in startup-sequence. This way scsi.device is patched.
- I'm using SFS for all partitions >4GB, so no reason to patch the filesystem
- I can only make the other (os3.9-formatted) partitions to show up using the diskchange command
- The problem is, hd2 for example shows up as a 3GB drive, while it really is 15GB in size.
- Also, HDToolbox does not display the correct size for the drive (strangely enough: 10GB in total ?)
- Oh, and when I start it, it asks me to save changes to disk (RDB) but Im afraid to lose everything when I do that.
- Check4gb displays all partitions (except those that are not on auto-mount) correctly. scsi.device v119.4, sfs2 v1.279, correct sizes.
- I think I've read all forum threads on this matter but the question remains:

How do I get (all) partitions to show up in the correct sizes under Workbench 3.1?

[save2600]

Someone might have a better suggestion, but...

For 3.1 to properly recognize a HD larger than 4gb, you'll need to add an OS3.5/3.9 type patch in your startup-sequence called 'New Style Device' or NSDPatch. Then you have to edit the NSDPatch.config file to reflect your hardware and SCSI or IDE setup. Here's the linky:

http://aminet.net/package/disk/misc/NSDPatch43_20

...and maybe there's something in here you could read up on:

http://aminet.net/package/disk/misc/SFS

[x303]

Hm, I don't understand you went back to wb3.1 in the first place.
Secondly, I don't think SFS2 partitions work under OS3.x. :confused:

x303

[Thomas]

- The problem is, hd2 for example shows up as a 3GB drive, while it really is 15GB in size.

The remainder of 15 / 4 is 3. Normal behaviour of old programs. There is nothing you can do except to use new programs.

For 3.1 to properly recognize a HD larger than 4gb, you'll need to add an OS3.5/3.9 type patch in your startup-sequence called 'New Style Device' or NSDPatch. Then you have to edit the NSDPatch.config file to reflect your hardware and SCSI or IDE setup. Here's the linky:

Complete nonsense. NSDPatch does nothing like this.

[rvo_nl]

Someone might have a better suggestion, but...
 
For 3.1 to properly recognize a HD larger than 4gb, you'll need to add an OS3.5/3.9 type patch in your startup-sequence called 'New Style Device' or NSDPatch. Then you have to edit the NSDPatch.config file to reflect your hardware and SCSI or IDE setup. Here's the linky:
 
http://aminet.net/package/disk/misc/NSDPatch43_20

thanks, is this patching the filesystem, scsi.device, or something else?

[rvo_nl]

Complete nonsense. NSDPatch does nothing like this

thanks, thats what I thought.
 

The remainder of 15 / 4 is 3. Normal behaviour of old programs. There is nothing you can do except to use new programs.

I dont understand what you mean by this. Im talking about workbench here. Surely its not safe to write to a partition which size is incorrectly shown?

[Thomas]

Secondly, I don't think SFS2 partitions work under OS3.x.

They do. You need SFS2 to have partitions bigger than 128GB. Even files bigger than 4GB are supported, if the programs support them.

is this patching the filesystem, scsi.device, or something else?

It patches scsi.device. It does not help with partition sizes because scsi.device does not know about partitions.

[save2600]

Complete nonsense. NSDPatch does nothing like this.

 I've woken up many a large GB hard drive with that patch. OR, am I confusing it with TD64??

http://aminet.net/package/disk/misc/td64patch

I'd normally cry foul here, but Thomas you are right 99.99% of the time. What "new programs" are you talking about in order for this guys HD to be recognized properly? Newer OS's you mean? OS3.9 for example wouldn't be choking this much on the sizes, but 3.1 sure does. And I'm 98% certain I used one of the two aforementioned patches in order to wake up the other partitions... I have a 3.1 partition on my 3.9 native system as a matter of fact and everything works as it should since playing around with one of those patches (unless my memory is completely pear shaped right about now). lol  Sorry to the OP for the confusion if that's the case!!

[rvo_nl]

Hm, I don't understand you went back to wb3.1 in the first place.
Secondly, I don't think SFS2 partitions work under OS3.x. :confused:
 
x303

I didnt go back to 3.1, I just wanted a fast backup partition from which I can load either OS3.9 or OS4.0, which are both located after and sized bigger than 4GB.

[Thomas]

I dont understand what you mean by this.

Imagine one of those old calculators which had a fixed number of digits (8 usually). Now try to calculate 30000000 * 512 and see what the result is when stored in only 8 digits.

Im talking about workbench here. Surely its not safe to write to a partition which size is incorrectly shown?

It's completely safe. The problem is cosmetic only. It's Workbench which is the old program. The file system behaves perfectly ok, there is no danger for your files.

[rvo_nl]

okay, so, just to summarize what you've said: mounting the partitions manually with diskchange is expected and normal behaviour, and the incorrect filesize can be ignored when using workbench3.1. I do not need other patches and can safely write files to all partitions. I just shouldnt touch hdtoolbox in wb3.1 and I shouldnt format anything with it either. correct?

[Thomas]

What "new programs" are you talking about in order for this guys HD to be recognized properly?

There is nothing wrong with his HDD. He uses IDEfix and SFS, both support NSD and TD64 already, so the combination is ok, no additional patches needed.

The problem is Workbench. It multiplies the number of free blocks by the block size (giving the number of free bytes) and stores the result in a 32 bit number, ignoring the overflow. More than 4 billion does not fit into 32 bits.

HDToolbox does the same.  But that's nothing to worry about. It's only the display which is wrong. Internally it works correctly.

[save2600]

It's only the display which is wrong. Internally it works correctly.

Ahhh... this knowledge would have saved me hours/days/weeks trying to get the display to work correctly! I had always thought this scenario meant the OS was not handling the newer drives properly. And it did seem as if when writing to the HD (past the ranges that are incorrectly displayed), something would muck up, but perhaps that's another issue altogether. As always, thanks for the expertise and education Thomas!

[Thomas]

okay, so, just to summarize what you've said: mounting the partitions manually with diskchange is expected and normal behaviour

Sorry, I overlooked this:

- I can only make the other (os3.9-formatted) partitions to show up using the diskchange command

The problem is that all partitions are mounted before IDEfix is run. So SFS checks if scsi.device can read the partition and fails, so it makes them invisible. Your diskchange *after* IDEfix has been run does the check again and this time it succeeds.

It's completely ok to run diskchange this way, if you can live with it. You can add the diskchange commands to user-startup so that you no longer recognise them.

Another solution is to replace c:idefix in startup-sequence by c:loadide reset. By this IDEfix will be loaded resident and a reset will make it active already when partitions are mounted. But this needs an additional reset, just like OS 3.9 or OS 4.

[rvo_nl]

Thomas, again I am very thankful for your replies. Like save2600 said, your answers really save me lots of time. But most of all, I'm learning a lot from them. Hope I can return the favor to other people some day.
 
edit: wanted to add one thing, its only the free space count that seems to be incorrect. The '% full' and 'in use' indication of the partition seem to work alright. I tried filling it up with some big files, and after some copies it says:
 
21% full, 4MB free, 2160MB in use
 
when I copied over another 80MB it says:
 
22% full, 96MB free, 2240MB in use
 
I will continue to fill up 2 partitions after the first 4GB and in case my HD0 gets destroyed will report back here

edit 2: no wait, its only the percentage indicator that works.