FSV - Recover Deleted Files
		===========================

			Users' Guide
			============

 On RSX, when a file is created a file header is created for it in the
 Index  File.  When  the file is deleted the header remains unchanged,
 but  is de-allocated. Thus, for a period of time (until the header is
 re-used  or  the  blocks  the  file  occupies  are  re-used),  it  is
 theoretically  possible to recover the file by finding the header and
 hence  the  allocation  pointers  for the file. This can be useful in
 either of two cases:-

  (i)  Some    programs  are  accidentally  deleted  and  need  to  be
       recovered.

 (ii)  The    file    structure    becomes    corrupt    making   them
       inaccessible.

 In  either  case,  if  FILSAV  is  run  immediately, it is  possible,
 usually, to recover the files.

 The program can be run in any of the valid RSX ways:-

       RUN $FSV
 or    FSV
 or    FSV command-line

 The first two will result in a prompt of

       FSV>

 which will re-prompt after each filename until CTRL-Z is typed.

 Note  that  before  running,  LUN  1 must be assigned to the drive on
 which  the  input  disk  is loaded (but not mounted) and LUN 2 to the
 output disk for the new files. Thus, if the first format is used, the
 relevant LUN assignments must be built into it at taskbuild time.

 FILSAV will accept a command-file as input.

 There are only two legal lines that can be specified to FSV:-

 1.   Specify starting block for File Header Search

      FSV>nnnnnn

      where nnnnnn is the octal number of the first block you want FSV
      to  read  when searching for file-headers. This can be useful if
      you know that all file-headers sit in a certain part of the disk.


 2.   Specify a filename to be restored

      FSV> filespec

      where  filespec  defines  the  file  to  be  recovered. The only
      relevant  parts  of this are the filename, extension and version
      number.  If a UIC is specified it defines the UIC for the output
      file,  but  is  ignored as far as the input file is concerned as
      the  UIC  in the file header is the Owner UIC, not the UIC under
      which the file is stored. Obviously no device may be specified.

      If no version number is given, FSV lists out each version of the
      file  encountered  together  with the number of blocks, and then
      selects the one with the highest number of blocks.

 FSV may report the following error message:-

     NO SUCH FILE - if  no file header could be found corresponding to
                    the filespec given.

		FSV - Recover Deleted Files
		===========================

			System Guide
			============

 The  task is very simple in concept. It assumes that the input device
 (the  one on which the 'deleted' or 'lost' files reside), is assigned
 to  LUN 1, and that the output device (the one on which the files are
 to  be  put),  is  assigned to LUN 2. It then reads the input disk 90
 blocks  at a time and searches each block to see if it is the missing
 file header for the required file (it does this simply by checking to
 see  if  the  block  contains  the  relevant  filename, extension and
 version (if specified) in the correct format in the right place - the
 odds  against  any  other  block containing them is very high). If no
 version  number is specified it searches for each one and selects the
 one  with the most blocks allocated.

 In  either  case  it  then  creates  an output file on LUN 2 from the
 information in the blocks pointed to by the retrieval pointers in the
 header just found.

 FSV  is a non-privileged task with a taskname of ...FSV, and uses six
 units  (well 1, 2, 5, & 6 anyway), of which unit 1 should be assigned
 to  SY: and unit 2 to LB:, and should be built with the macro library
 [1,1]MACLIB.MLB  (for  $BASTS) and the object library [1,1]OBJLIB.OLB
 (for RADASC & suchlike).

 As  it  stands  FSV is horribly primitive (but appears to work).  The
 reason it searches the whole disk for the file header rather than the
 Index  File only is a fear that the Index File may be clobbered - but
 it  really  should  do  either  dependent  on a switch.  Also, at the
 moment,  the  disk  with  the  deleted  file  must  be  taken offline
 immediately - which is awkward in a muti-user environment.  Maybe one
 day I'll tidy it up a bit!