DOS ain't dead

Hello,

as some of you are probably aware of, DOSLFN 0.40e from Jason Hood has (a) serious bug(s) that may cause corruption of directory entries.

I was able to track (one of) the bug(s) down. After the bug was fixed, I didn't experience corruptions anymore as of yet, but there is no guarantee. That's why I post the fix here: you should try it and report your experiences.

Also, the fix is for the source code only, because DOSLFN is copyrighted software - you will have to assemble and link the source.

1. change DOSLFN.ASM

@@pop_open_only:
        pop     ax
if 1
;--- japheth: BX is not set to a DIRENT here!
;--- so we cant copy anything to/from [bx] as it's done below.
        call Copy_FCB_8P3_from_FCB_to_DI
        jmp SFN_6C_CallOld      ;just call old int 21h, ax=6c00h
endif
@@open_only:


2. assemble and link DOSLFN.ASM to DOSLFN.COM:

tasm -m5 -la doslfn.asm
tlink -3 -t -s doslfn.obj


---
MS-DOS forever!

> Also, the fix is for the source code only, because DOSLFN is copyrighted
> software - you will have to assemble and link the source.

Isn't there something about that in the documentation or source? In any case, when I asked Henrik some time ago about some other source code he provided on his website, he told me that he basically doesn't care what people do with it, he only asks for attribution.

---
l

> as some of you are probably aware of, DOSLFN 0.40e from Jason Hood has (a)
> serious bug(s) that may cause corruption of directory entries.
>
> I was able to track (one of) the bug(s) down. After the bug was fixed, I
> didn't experience corruptions anymore as of yet, but there is no guarantee.
> That's why I post the fix here: you should try it and report your
> experiences.

Have you reported this to Jason directly? Would be better for him to fix his version.

> Also, the fix is for the source code only, because DOSLFN is copyrighted
> software - you will have to assemble and link the source.

A context diff would've been nicer here (and I'm pretty sure copyright law allows that).

But yeah, anyways, thanks, I'll try to try it out.

> > I was able to track (one of) the bug(s) down.

Apparently there's another bug (regression) discovered by Juan on comp.os.msdos.djgpp (re: DJGPP 2.04 w/ MS-DOS 6.22 + DOSLFN 0.40e):

> I have found my old FreeDOS CD and installed DOSLFN 0.40c.
> With this driver everthing works on MSDOS. Stepping into
> the filelength() code shows that the 0.40e version does
> not set the CF when it returns from 0x71A6 call while 0.40c
> does so that it jumps directly into the 0x4201 code instead
> of returning an EOVERFLOW.

> Apparently there's another bug (regression) discovered by Juan on
> comp.os.msdos.djgpp
> (re: DJGPP 2.04 w/ MS-DOS 6.22 + DOSLFN 0.40e):
>
> > I have found my old FreeDOS CD and installed DOSLFN 0.40c.
> > With this driver everthing works on MSDOS. Stepping into
> > the filelength() code shows that the 0.40e version does
> > not set the CF when it returns from 0x71A6 call while 0.40c
> > does so that it jumps directly into the 0x4201 code instead
> > of returning an EOVERFLOW.

I don't think this is a bug in DOSLFN, it is almost certainly a bug in DPGJJ - as usual.

Why? Because it's known and old hat that DOSLFN does NOT support function 71A6h:

DOSLFN.ASM itself:


verteiler:      DVT     39h,lfn_mkdir   ;w DS:DX
                DVT     3Ah,lfn_rmdir   ;w DS:DX
                DVT     3Bh,lfn_chdir   ;r DS:DX
                DVT     41h,lfn_unlink  ;w DS:DX       CX SI
                DVT     43h,lfn_attr    ;? DS:DX BL    CX SI DI
                DVT     47h,lfn_pwd     ;r DS:SI DL
                DVT     4Eh,lfn_ffirst  ;r DS:DX ES:DI CX SI
                DVT     4Fh,lfn_fnext   ;r BX    ES:DI    SI
                DVT     56h,lfn_move    ;w DS:DX ES:DI
                DVT     60h,lfn_name    ;r DS:SI ES:DI CX
                DVT     6Ch,lfn_creat   ;? DS:SI    BX CX DX DI
                DVT    0A0h,lfn_volinfo ;- DS:DX ES:DI BX CX DX
                DVT    0A1h,lfn_fclose  ;r BX
;--- entry 0A6h ( get file info by handle ) is missing
                DVT    0A7h,lfn_timeconv;- DS:SI       BX CX DX
                DVT    0A8h,lfn_genshort;- DS:SI ES:DI DX
                DVT    0AAh,lfn_subst   ;- DS:DX BH(=0,1,2) BL(=LW)
                db      0


or DOSLFN.TXT

THAT WILL PROBABLY NEVER WORK:

* functions around SUBST, AL=AAh        ("query subst" is available)
* file creation from server, AL=A9h
* retrieve handle information, AL=A6h
* reset drive, AL=0Dh


That's why calling this function must be done very carefully. See HX, DKRNL32, file GETFINFO.ASM:


        mov ebx, handle
        mov edx, pFileInfo
        mov ax,71A6h            ;this is NOT implemented in DOSLFN!!!
        stc
        int 21h
        jnc exit
        cmp ax,7100h
        jnz error


And my suspicion is that DGVJJ does NOT call this function "very carefully".

---
MS-DOS forever!

> Why? Because it's known and old hat that DOSLFN does NOT support function 71A6

King Udo supports it ... but it's buggy

---
This is a LOGITECH mouse driver, but some software expect here
the following string:*** This is Copyright 1983 Microsoft ***

> And my suspicion is that DGVJJ does NOT call this function "very carefully".

They just brewed a path for over 10 years old code:

--- djgpp.orig/src/libc/posix/sys/stat/lfilelen.c 2001-02-04 19:13:00 +0000
+++ djgpp/src/libc/posix/sys/stat/lfilelen.c 2011-09-04 12:25:38 +0000

(link above)

---
This is a LOGITECH mouse driver, but some software expect here
the following string:*** This is Copyright 1983 Microsoft ***

Today I compiled your patch with TASM 4.1 and it worked without any problems. Thanks Japheth!

I think it is very important patch which could be applied in FreeDOS 1.1.

---
DOS-u-akbar!

> Today I compiled your patch with TASM 4.1 and it worked without any
> problems. Thanks Japheth!

Please could you upload your binaries somewhere (or send me via mail)? Thx

---
DOS gives me freedom to unlimited HW access.

> > Today I compiled your patch with TASM 4.1 and it worked without any
> > problems. Thanks Japheth!
>
> Please could you upload your binaries somewhere (or send me via mail)? Thx

doslf40f.zip

> doslf40f.zip

Thx, I already got fixed ver from Laaca...

---
DOS gives me freedom to unlimited HW access.

> >
> doslf40f.zip
>
> Thx, I already got fixed ver from Laaca...

i have here a doslfn which seems to be a continuation of version 0.40E

Version 0.40F
+ Better support for DBCS code pages and more

http://www.multiupload com/7OTZTRTHFD

>
> Version 0.40F
> + Better support for DBCS code pages and more
>
> http://www.multiupload com/7OTZTRTHFD


another doslfn "F"

multiupload.com/YTZ1JV7CAF

> >
> > Version 0.40F
> > + Better support for DBCS code pages and more
> >
> > http://www.multiupload com/7OTZTRTHFD
>
>
> another doslfn "F"
>
> multiupload.com/YTZ1JV7CAF

Two different files, hmmmm, weird. But yeah, I remember at least one of these. IIRC, the only problem (confirmed as still) is that both are closed source! Argh. So that's fairly useless.