Path: utzoo!utgpu!news-server.csri.toronto.edu!bonnie.concordia.ca!thunder.mcrcim.mcgill.edu!snorkelwacker.mit.edu!usc!elroy.jpl.nasa.gov!mahendo!nereid!john
From: john@nereid.jpl.nasa.gov (John Veregge)
Newsgroups: comp.sys.amiga.datacomm
Subject: Re: ftp->mtools->d2d->Zoo->corrupt files
Keywords: ftp mtools d2d zoo sun sparc
Message-ID: <6060@mahendo.Jpl.Nasa.Gov>
Date: 24 Jan 91 16:59:23 GMT
References: <6057@mahendo.Jpl.Nasa.Gov>
Sender: news@mahendo.Jpl.Nasa.Gov
Reply-To: john@nereid.jpl.nasa.gov (John Veregge)
Organization: Technology Development Group (JPL)
Lines: 54

In article <6057@mahendo.Jpl.Nasa.Gov>, john@nereid.jpl.nasa.gov (John Veregge) writes:
|> 1. Ftp fish disks from gatekeeper.dec.com to a Sun Sparc station.
|>    (The files are zoo format, that is *.zoo, in /pub/micro/amiga/fish.)
|> 2. Format standard amiga floppy disk on the Sun with 'mkdfs -f'.
|>    (Format disk as a Sun Unix disk and then lay the MSDOS filesystem on top.)
|> 3. Copy Unix files to amiga floppy with 'mcopy *.zoo a:'
|> 4. Copy MSDOS file to Amiga with Dos2Dos.
|> 5. Unarchive files with 'zoo -extract filename'.
|> 6. Zoo begins unarchiving, but always exits prior to completion with
|>    'CRC error, file probably corrupted'. (Zoo is version 2.0)
|> 
|> With a straight text file (and the approprite command line switches) this
|> download method works fine, leading me to guess that the problem is with zoo.
|> 

So far all responses have suggested that I misused ftp (ascii vs binary). One
soul was kind enough to mail me a copy of zoo to check (Phil Kernick). I didn't
make that mistake (although paranoia required I use the zoo to verify what I
already knew). Another suggestion was to check the disk in a MSDOS machine,
however Dos2Dos reads MSDOS disks and had no problem recognizing the files.

The problem is in MTOOLS. I was able to use the unix zoo in a simple test:
zoo -test file -> mcopy file a: -> mcopy a:file . -> zoo -test file
The file on return was corrupted. The file corruption occurred with both a
disk formatted by Dos2Dos and a disk formatted by MTOOLS' mkdfs. Therefore
the problem is probably in mcopy or something mcopy calls (mread or mwrite).

I recently built MTOOLS on the Sun Sparc Station 1, using the Sun supplied
compiler, in OS version 4.1.0. The only problem I had compiling was in the
function formatit(), in file mkdfs.c, but the problem was in error reporting
only (Something had changed in the wait(2) function and/or it's data
structures.) and I just commented it out. Below is the commented out code.

#if 0
  if ( WIFSIGNALED(w) ) {
    fprintf(stderr,"%s: ",progname) ;
    psignal(w.w_termsig,"/bin/fdformat") ;
    if (w.w_coredump)
      fprintf(stderr,"Core dumped\n") ;
    exit(1) ;
  }

  assert( WIFEXITED(w) ) ; /* If not signalled,  must be exited
                              we are not tracing stopped processes */
#endif

Any other suggestions?

-- 
John R Veregge                 Section 348 - Flight Command and Data
Jet Propulsion Laboratory      Management (Technology Development)
Calif Institute of Technology  Mail stop: T1704, Office: T1704-P
4800 Oak Grove Drive           Phone: (818) 354-0511, FAX: 393-4494
Pasadena, CA, USA 91109        john@triton.jpl.nasa.gov