Creating a SFS HDF file

[Bloodwych]

I'm trying to create an SFS HDF in WinUAE, but can't get it to work. I need to do this to test some ClassicWB extras I've added to the popup menu.

I've tried the following:

Adding the SFS filesystem when creating a HDF. Entered workbench (which has SFS in the L directory) and used SFSformat and format. Both cause an indefinite pause on initializing disk.

Adding a HDF, clearing all parameters accept blocksize and attempted to run HDtoolbox with the "device=uaehf.device" parameter. No disks show up.

What am I doing wrong? OS is Win ME (does uaehf.device only work under 2k/xp), WinUAE version 0825 (waiting for 1.0 before jumping fully to the new GUI).

[Toni Wilen]

Regular HDF: you need to use hex-editor to add "SFS\0" identification to beginning of HDF-file. (because WinUAE won't load the filesystem if it is not needed..)

Full HDF: try removing unneeded directory or hardfile drives (or move the HDF higher in harddrive config panel) because older HDToolBox versions won't detect drives with "too large" unit number.

[Bloodwych]

Thanks Toni, excellent advice.

[Bloodwych]

Only I can't make use of it!

When I use HDtoolbox FullHDF method and set everything up for SFS, it always says "save failed, not enough room for configuration data"

I tried editing a regular HDF file with SFS\0 identifier at the beginning, but can't get this second method working either. Same problem as before.

Any ideas what might be wrong? I'm determined to mount a SFS partition in WinUAE.

Although I'd rather do this myself, if someone could drop a zipped empty SFS HDF in the zone (at least 100MB when unzipped) I'd be most grateful!

[Toni Wilen]

Originally Posted by Bloodwych

When I use HDtoolbox FullHDF method and set everything up for SFS, it always says "save failed, not enough room for configuration data"

This means there is not enough space in RDB for all filesystems. Try deleting FastFileSystem before adding SFS.

I tried editing a regular HDF file with SFS\0 identifier at the beginning, but can't get this second method working either. Same problem as before.

Did you use "SFS<binary zero>", not "SFS<ascii zero">? Do you see something about fake filesystem in winuaelog.txt?

for example:

hardfile2=rw,DHX:D:\amiga\plain.hdf,32,1,2,512,0,D:\amiga\AMIGA\l\SmartFilesystem

Beginning of plain.hdf:

00000000h: 53 46 53 00 00 00 00 00 00 00 00 00 00 00 00 00 ; SFS.............
00000010h: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ; ................

[Bloodwych]

Sorry for taking your time Toni, really appreciate the help. I know you must have better things to do - like WinUAE development.

Already tried deleting FFS using the "add/update" filesystem in HDtoolbox. Then after "adding" SFS, using the correct identifier (0x53465300) and selecting SFS/0, I still get that message when attempting to save. I'll go back and check I'm not doing anything stupid then post a detailed explanation of the steps I've taken.

Tried that SFS at the beginning - like before it formats then fails to initialize using SFSformat command and the normal workbench format (see pic below). The HDF after a failed format now contains this at the beginning:

00000000h: 42 41 44 00 53 46 53 01 53 46 53 02 53 46 53 03; BAD.SFS.SFS.SFS

WinUAE config has this after I've setup the HDF:

hardfile2=rw,DH0:G:\Test.hdf,32,1,2,512,0,D:\Emulators\WinUAE0825\SmartFilesystem
hardfile=rw,32,1,2,512,G:\Test.hdf

WinUAE Boot only has this:

hfd open: 'G:\Test.hdf'
HDF 'G:\Test.hdf' opened successfully, handle=00000078

[Toni Wilen]

Originally Posted by Bloodwych

WinUAE Boot only has this:

hfd open: 'G:\Test.hdf'
HDF 'G:\Test.hdf' opened successfully, handle=00000078

winuaelog.txt, not winuaebootlog.txt. (and don't forget to enable logging)

Part of my winuaelog.txt:

Fastmem (32bit): mapped @$10000000: 64 MB Zorro III fast memory
Card 1 (ZorroIII) done.
UAEGFX-card: mapped @$14000000
Card 2 (ZorroIII) done.
Filesystem: mapped memory @$e90000.
Card 3 (ZorroII) done.
Catweasel MK3: mapped @$ea0000
Card 4 (ZorroII) done.
FS: mounted virtual unit DH0
FS: mounted HDF unit DHX
RDB: fakefilesys, trying to load 'D:\amiga\AMIGA\l\SmartFilesystem', dostype 0x53465300
HDF: faked RDB filesystem 53465300 loaded

(DHX is my test SFS hardfile, DH0 is regular directory harddrive)

HDToolbox problems: sorry, no idea. It works fine here

[Bloodwych]

Woops, sorry! Getting late in the day.

winuaelog:

Fastcard: mapped @$200000: 8MB fast memory
Card 1 (ZorroII) done.
Filesystem: mapped memory @$e90000.
Card 2 (ZorroII) done.
FS: mounted HDF unit DH0
RDB: fakefilesys, trying to load 'D:\Emulators\WinUAE0825\SmartFilesystem', dostype 0x53465300
HDF: faked RDB filesystem 53465300 loaded

I'll try copying your config.

If it works your end, I guess we're at a stop with HDtoolbox. Like I said I'll test further.

[Bloodwych]

Well, I've tried and failed. Both methods won't work. SFS wasn't meant to be.

Thanks for your help.

As a last desperate step incase you see any stupid error, I give caps of my SFS nightmare for the Full HDF method......

[Bloodwych]

....continued.

[Toni Wilen]

Hmm.. HDToolBox version?
("version tools/hdtoolbox")

I have used newer HDToolBox that comes with OS3.9. Maybe some older versions have some compatibility problems with WinUAE or SFS?

[thomas]

Well, you entered 68 blocks per cylinder. This is 68 KB for the first two cylinders. Very little room for an RDB + file system.

You should use about 1000 blocks per cylinder, so there is at least one MB for the RDB.

For an 80MB HDF I'd suggest to use 2 Heads, 256 Sectors and 320 Cylinders.

[Bloodwych]

Thank you Thomas! Once again the Amiga guru lets his advice save the day.

And thank you Toni for sticking with me in this thread, much appreciated. Sometimes it's hard to determine what silly mistake the other person has made. Good job I eventually posted those pics! I think I'll write a guide for the FAQ section.

The answer was so obvious! I can't believe I didn't think of changing the physical characteristics - after all it's just a virtual file, I change them to whatever I want! How I failed to check what size the RDB was being physically defined as is worrying.

Oh well, these things happen.

I didn't actually enter those details Thomas. In fact I didn't even notice them. I'm so used to dealing with real HDs and fixed physical characteristics, I simply used the read config button which set silly values that I failed to notice. I took one look at the 79MB and thought everything was fine. However, HDtoolbox can't determine block size correctly with HDFs.

Oh well, this info is here for anyone else in the future.

[Toni Wilen]

Drive geometry is always faked except if you have XT or MFM drives

thomas: does newer HDToolbox-versions "invent" different geometries because I haven't seen this problem with OS3.5+ HDToolBox versions?

btw, regular SFS hardfiles still work fine here. Could you try testing with WinUAE 0.9.90? (released very soon)

[thomas]

does newer HDToolbox-versions "invent" different geometries because I haven't seen this problem with OS3.5+ HDToolBox versions?

The only thing I know is that HDToolbox was totally rewritten in OS3.5. It now uses hdwrench.library which was invented with OS3.5. I never found a documentation on how the logical drive geometry is calculated. However, I believe that the new HDToolbox only uses the total number of blocks and calculates all the other values by itself. Maybe older versions also use TD_GETGEOMETRY to get the "real" geometry.

One obvious difference is that the 3.1 HDToolbox always uses 1008 blocks per cylinder for IDE drives and with 3.5 it uses 2016 blocks. Also if you partition a 100MB ZIP disk, I believe that with 3.1 it uses 2 heads, 64 sectors and 1536 cylinders (what I consider the real geometry of the disk) while 3.5 uses 16 heads, 48 sectors and 256 cyls.

[D-Dan]

Originally Posted by Toni Wilen

btw, regular SFS hardfiles still work fine here. Could you try testing with WinUAE 0.9.90? (released very soon)

I've been running my Amiga system on WinUAE for the past 3 - 4 versions, formatted with SFS and no problems. believe me, if 0.9.90 has any problems with SFS I'll let you know - loudly

[Bloodwych]

Thomas: Well as I said before, the older HDtoolbox only used 68 Blocks per cylinder/track when selecting "Read Config" with a cleared HDF. It says it can't detect the value, so "assumes" 68.

Toni: Yes I'll give the regular HDF method another try once I get my hands on the new WinUAE.

D-Dan: Are you using the Regular HDF method? Any tips or advice?

[Toni Wilen]

SFS hardfile problem solved!

WinUAE 0.8.27 and older: Only SFS older than 1.187 work. 1.187 removed SCSI-direct and requires TD_GETGEOMETRY but only WinUAE 0.9.90 has hardfile TD_GETGEOMETRY-support..

This does not affect RDB-hardfiles.

[Bloodwych]

Brilliant work Toni!

Next release is going to be sweet. I may jump ship before the 1.0 release after all these changes.

[D-Dan]

Originally Posted by Bloodwych

D-Dan: Are you using the Regular HDF method? Any tips or advice?

Sorry - didn't see your post 'til this morning, so maybe a moot point now, but I just set the hardfile as normal, fired up HDToolBox in UAE (OS3.9 version), installed SFS to the RDB, and formatted.

Exactly the same as prepping a real HD on Amiga, and it worked without a problem.

This is with the latest SFS.

Sorry I can't give more info.