Anyone working on PFS3 ??

[Piru]

@Franko

You asked for proof and I provided it. Care to explain how my proof is incorrect?

[Piru]

The only problems you will encounter are application displaying the wrong file size as most were written never expecting to take files of >4GB into account...

With AmigaOS 3.x that would be all applications. APIs in AmigaOS 3.x can only represent a 32bit number for the file size.

Only applications reading up until EOF will work properly. Any AmigaOS 3.x application that actually queries the file size first before processing that amount of data will fail. Many applications seeking in a file will fail randomly as negative return value from Seek() was commonly considered an error.

Oh I'm sorry for giving you more technically correct and factual information. You may yell "YOU'RE WRONG!" if that makes you feel any better.

[Piru]

You'd actually need 43 bits to represent that value (unsigned). Logarithms FTW

Actually 33 bits. TD_GETGEOMETRY and RDB use block offsets, not byte offsets.

[Piru]

MorphOS: Yes, Harry is working on that.
AROS: ?
OS4: ?
OS3: ?

Currently almost all changes are generic. Of course the OS3 version is directly benefiting from the changes being made. There's only one SVN tree that has everything. Assuming the file system is ported to AROS and OS4 it'd be highly desirable to have those changes in the same SVN as well.

Anyone wishing to contribute should contact mr Pelt over sourceforge.

[Piru]

So what compiler are you using ?

ppc-morphos-gcc and SAS/C 6.58

I've tried to compile it myself with SAS/C and VBCC

VBCC is unlikely to work. Currently things build with SAS/C and gcc.

but it only gives me lots of errors of non existing items (IPTR, DOSBase->dl_Root [dl_root only exists in dos/dosextens.i, not .h] ,

It builds with SAS/C 6.58 and NDK 3.9 includes at least:
http://www.haage-partner.de/download/AmigaOS/NDK39.lha

gadgets/progress.h missing).

gadgets/progress.h is in pfs-amiga/PFSDoctor/trunk. If you don't want to copy it you can modify the pfs-amiga/PFSDoctor/SCOPTIONS file and add: INCDIR=include: INCDIR=include

[Piru]

Well, in pfs3.h there's a line which says: '#if defined(__GNUC__) || defined(__VBCC__)'. So I thought it might also be build on VBCC.

Well, I don't know if it does. All I know is that it does build with gcc and SAS/C.

Strangely DOSBase->dl_root is in dosextens.h from NDK3.9, but not in the newer version v 1.170 that says 'Copyright (C) 1985-2005 Hyperion Entertainment VOF and Amiga, Inc.'

I see, Hyperion is at it again breaking legacy compatibility. I'd recommend you do NOT use OS4 includes then.

No. It's not there. Please add.

Hmm, I guess I have it there locally then. There are some uncommitted changes in the 68k build since I didn't want to touch those in too many ways.

The official build instructions are in the README file. I think you'd better follow those instructions.

[Piru]

@Cosmos

Use the bug tracker, please. Also, please include output from check4gb.

[Piru]

@Piru

This new version of Pfs3 give uninitialised partitions beyond the 4 Gb border : "Write Error -3 on block 0" when used with the scsi 43.45 (Boing Bag 2) burned on eprom...

For whatever reason this device doesn't appear to have a working HD_SCSICMD interface (DirectSCSI). This is indicated by the Check4GB output as well (no S).

This doesn't explain why old PFS3 supposedly works. I suspect some problem with your testing method rather than that the old PFS3 would actually be working.

[Piru]

Unbelievable... The v18.3 is working fine, YOU made some changes and bump to v18.4...

I did not bump the filesystem version.

Code: [Select]

    2   michielp #define REVISION "18.4"
     2   michielp #define REVDATE  "11.09.99"
     2   michielp #define REVTIME  "18:02:53"
     2   michielp #define AUTHOR   "Michiel Pelt"
     2   michielp #define VERNUM   18
     2   michielp #define REVNUM   4


v18.4 give now partitions uninitialised beyond the 4 Gb border, and it's ME the culprit...

v18.5 is the official PFS3 without any of my changes. If it fails then I can't see how this could be my fault.

Finally: The svn version is a work in progress. Anyone using this snapshot is doing so with the knowledge that things may be in flux. I strongly suggest you avoid using random svn builds and rather use the official v18.5 for now.

[Piru]

used with the scsi 43.45 (Boing Bag 2)

Really? Isn't that 43.45 some unofficial patch from aminet?
http://aminet.net/package/driver/media/SCSI4345p

Code: [Select]

THIS PATCH IS COMPLETELY UNOFFICIAL. IT HAS NOT BEEN APPROVED BY AMIGA INC.
OR BY HAAGE & PARTNER. BY USING IT ON YOUR SYSTEM, YOU ACCEPT THE ENTIRE
RISK AS TO ITS QUALITY AND PERFORMANCE.

[Piru]

When coders add some new changes on some sources, they bump the version or the revision numbers...

The source committed to SVN was for earlier PFS3 version (iirc PFS3 5.2 aka 18.4), since the very last src were lost. The source code has since been brought up to date and the current SVN should match the PFS3 5.3 (18.5) in features. See here: http://sourceforge.net/tracker/?func=detail&aid=3285730&group_id=532591&atid=2163218

The SVN "trunk" version is work in progress (and as such the revision hasn't been touched yet). The SVN trunk version isn't guaranteed to be stable at any given moment. Obviously there's nothing we can do to prevent anyone from releasing snapshot builds, but I would appreciate it if any such release would be accompanied with a warning about it being only a snapshot build.

[Piru]

Pfs3 versions uploaded on Aminet in 2011 :

- binaries => v18.3
- sources => v18.4

http://aminet.net/package/disk/misc/PFS3_53 contains v18.5 though.