Google
 

Trailing-Edge - PDP-10 Archives - bb-bt99e-bb - utl702.d11
There are no other files named utl702.d11 in the archive.
                 EDIT DESCRIPTIONS FOR UTILITIES-10-V702  
  
                             EDIT 522    FOR BACKUP
  
[SYMPTOM]
  
     If the INITIAL file specification  is  not  found,  or  contains  a
non-existent  path,  the  save  will  start from an unpredictable place,
rather than not saving anything (the correct case).
  
[DIAGNOSIS]
  
     The INITIAL code assumes that the file in question  always  exists,
and  exists  on  the  first file structure searched if DSK:  or ALL:  is
specified.  If only part of the path is found, or  the  entire  path  is
found,  but  not  the correct filename, any subsequent path and filename
which satisfies the constraints  remaining  from  the  previous  (nearly
correct) attempt will succeed.
  
[CURE]
  
     Make  sure  that  the entire path and filename match on any INITIAL
file.  If the path or file is not found, abort the search on the current
file structure, and move on to the next structure.
********************************************************************************
  
  
                             EDIT 523    FOR BACKUP
  
[SYMPTOM]
  
     If the INITIAL file is not found during a  SAVE,  BACKUP  types  an
error message to the effect that no files exist anywhere.
  
[DIAGNOSIS]
  
     All BACKUP knows is that no files were saved or restored.  It  thus
types  a no files found error message for all the filespecs given on the
SAVE/RESTORE command.
  
[CURE]
  
     Before  saying  nothing  was  found,  look  for the existence of an
unsatisfied INITIAL filespec.  If this  is  the  case,  issue  an  error
message stating that the INITIAL file was not found.
********************************************************************************
  
  
                             EDIT 524    FOR BACKUP
  
SYMPTOM
  
  
DIAGNOSIS
  
  
********************************************************************************
  
  
                             EDIT 525    FOR BACKUP
  
[SYMPTOM]
  
A BACKUP RESTORE gets "Prior block repeated" warning  at  the  end  of
tape if the tape was mounted unavailable when SAVEd.  This message did
not exist for versions of BACKUP prior to version 3.   However,  there
were still extraneous records written at end-of-tape.
  
[DIAGNOSIS]
  
For purposes of this discussion, define a buffer ring with buffers  A,
B,  C,  and D where A points to B, B points to C, C points to D, and D
points to A.  This buffer ring is used for magtape output in BACKUP.
  
When BACKUP gets an error return  doing  an  OUT  UUO  because  of  an
end-of-tape condition, it tries to force output of all buffers not yet
written to the tape before it writes an end-of-volume record,  does  a
reel  switch,  and writes the necessary records for the start of a new
tape.  When this error return happens, the buffer  ring  header  block
points  to  buffer  A.   Buffer A has the IO.EOT bit set in the buffer
header block, and buffers B, C, and D have the use bit set.   (When  a
use  bit is set, it means that the buffer is available for the emptier
of the buffer.) A SETSTS is done to clear the EOT  condition.   BACKUP
stores  the  address  of  the buffer which is pointed at by the buffer
ring header block to mark where the search for buffers not yet  output
will end.
  
BACKUP then steps through the ring searching for a buffer with the use
bit  set.   In  this  case,  BACKUP finds the use bit set in buffer B.
Buffer B's address is saved.  All the use bits are now cleared and  an
OUT  UUO  is done with the argument pointing at buffer B.  This output
succeeds and BACKUP checks to see if the current buffer address in the
buffer  ring  header  block  is  the  same as the buffer address where
BACKUP started (buffer A).  It  is not, so BACKUP adjusts  the  buffer
ring  header  block appropriately so that the next OUT UUO will output
buffer C.
  
Now an OUT UUO is done with the buffer ring header block  pointing  at
buffer  C.   This  OUT gets an error return with an EOT condition.  An
examination of the buffer ring shows that buffers A  and  B  now  have
IO.EOT  set  (buffer  A's  bit  was  never  cleared from the first OUT
error), the buffer ring header block points at buffer D  and  the  use
bit  is  set  in  buffer  C.  The routine FNDBUF is called to find the
buffer with IO.EOT set using the current buffer position (buffer D) as
a  starting  point.   FNDBUF returns erroneously pointing at buffer A.
IO.EOT is cleared in buffer A and a SETSTS is done to clear EOT.
  
Now all use bits are cleared in the buffer ring and an OUT UUO is done
with  the  argument  pointing  at buffer B.  THE TAPE NOW CONTAINS TWO
RECORDS OF BUFFER B which causes the "Prior block repeated" message to
occur on a restore.
  
BACKUP laboriously repeats  the  previous  two  paragraphs  until  the
address in the buffer ring header block is equal to the buffer address
which was stored when the first OUT UUO failed.  The reel switch  then
occurs and BACKUP continues the SAVE process.
  
This analysis shows that a BACKUP tape will contain redundant data but
no  missing  data.   This problem only happens if BACKUP does the reel
switch;  there is no detected problem when PULSAR does the reel switch
in BACKUP's behalf.
  
[CURE]
  
When the first OUT UUO error return occurs because of  an  end-of-tape
condition, clear IO.EOT in the buffer.
  
This may not be the most elegant solution, but  it  is  isolated  and,
therefore,  the safest for patching purposes.  BACKUP's magtape output
routine is too convoluted to streamline without extensive testing.
********************************************************************************
  
  
                             EDIT 451    FOR DIRECT
  
  
  
[SYMPTOM]
  
     The alignment of the device name is poor, when dir/sort/words  is
used.
  
  
  
                        **********
  
  
[DIAGNOSIS]
  
     When the version number is less than or equal  to  6  characters,
the  device  name  is  printed at the next tab stop but does not align
with the device name which is printed if the version number is greater
than or equals 7 characters.
  
     In the call to LDATEN for the date used with /SORT, one space  is
output before the date.  This  space  caused the improper alignment of
the device.
  
  
  
                        **********
  
  
[CURE]
  
     Allow all device names to be printed  in  the  same  column  when
DIRECT/SORT/WORDS  is  used.   Take  out  the space before the date is
output from the /SORT and add two spaces before the version is output.
  
  
********************************************************************************
  
  
                             EDIT 452    FOR DIRECT
  
  
  
[SYMPTOM]
  
     When the following is done, the directory output does not utilize
the full width of the terminal.
  
     .DIRECT/WIDTH/FAST/PRDEV/PRDIR/HDSDIR
  
  
  
                        **********
  
  
[DIAGNOSIS]
  
     If /HDSDIR is used in conjunction with /PRDEV/PRDIR/W/F,  do  not
allocate  spaces  in  lines for the device and directory fields.  When
using /W/F/PRDEV/PRDIR/HDSDIR, the device  and  directory  information
will be output by /HDSDIR.
  
  
  
                        **********
  
  
[CURE]
     Do  not  allocate  spaces  for  the  device  and  directory  when
/PRDEV/PRDIR is used in conjunction with /HDSDIR.
********************************************************************************
  
  
                             EDIT 453    FOR DIRECT
  
  
  
[SYMPTOM]
  
     DIRECT/MVOLUME switch is not compatible with the  default  GALAXY
USER-EOT mode.
  
  
  
                        **********
  
  
[DIAGNOSIS]
  
     When using DIRECT/MVOLUME you have to mount first tape  and  when
that  tape is at the EOT the user would receive a message to mount the
next tape and type C to continue or E to exit.
  
  
  
                        **********
  
  
[CURE]
  
     Add volume switching.
********************************************************************************
  
  
                             EDIT 454    FOR DIRECT
  
  
  
[SYMPTOM]
  
     The alignment of the device name is poor, when dir/sort/words  is
used.
  
  
  
                        **********
  
  
[DIAGNOSIS]
  
     When the version number is less than or equal  to  6  characters,
the  device  name  is  printed at the next tab stop but does not align
with the device name which is printed if the version number is greater
than or equals 7 characters.
  
     In the call to LDATEN for the date used with /SORT, one space  is
output before the date.  This  space  caused the improper alignment of
the device.
  
  
  
                        **********
  
  
[CURE]
  
     Allow all device names to be printed  in  the  same  column  when
DIRECT/SORT/WORDS  is  used.   Take  out  the space before the date is
output from the /SORT and add two spaces before the version is output.
  
  
********************************************************************************
  
  
                             EDIT 456    FOR DIRECT
  
[SYMPTOM]
  
  
  
  
DIRECT won't accept multiple-volume multiple-saveset  tapes  with  the
/MVOLUME switch.
  
[DIAGNOSIS]
  
Confusion between start and continuation of save sets.
  
  
[CURE]
  
Don't try to verify volume information on a start of saveset.
********************************************************************************
  
  
                             EDIT 462    FOR DIRECT
  
[SYMPTOM]
  
  
  
DRTAMW MESSAGE ISSUED TOO OFTEN FOR ESTHETICS
  
  
The "DRTAMW File attributes may be wrong" warning  message  is  issued
disturbingly often.
  
  
[DIAGNOSIS]
  
The DRTAMW warning is issued once for each file read from a BACKUP  or
DUMPER  (INTERCHANGE) tape if that file's attribute block (as recorded
on the tape) claims no attributes (or too few), yet seems to  actually
have the attributes on the tape.
  
  
[CURE]
  
Only issue the  warning  message  once  per  tape  save  set,  on  the
assumption  that  if  any  files  are "questionably" written, then all
files will be so written (and one message for each file  is  not  only
redundant but annoying as well).
********************************************************************************
  
  
                             EDIT 463    FOR DIRECT
  
  
  
[SYMPTOM]
  
  
  
  
DIRECT issues spurious  'Tape  mark'  warnings  during  processing  of
multi-volume  BACKUP  save  sets  with  /MVOLUME  switch  when tape is
MDA-controlled with USER-EOT..
  
  
                        **********
  
  
[DIAGNOSIS]
  
Logical 'EOF' is not cleared on an MDA /LABEL:USER reel switch.
  
  
                        **********
  
  
[CURE]
  
Clear IO.EOF (CLOSE monitor call) on all reel switches.
********************************************************************************
  
  
                             EDIT 465    FOR DIRECT
  
[SYMPTOM]
  
  
  
  
Checksummed listings may be wrong for BACKUP/FRS tapes  with  repeater
records.
  
  
[DIAGNOSIS]
  
BACKUP doesn't always light the GF$RPT this-is-a-repeater-record  flag
in  the  flags header word of repeater records.  DIRECT will therefore
treat the repeater record as a new valid record  and  checksum  it  as
well.   Further, the next record may not even be read in yet, in which
case DIRECT will ignore the possibility  of  there  being  a  repeater
record.
  
  
[CURE]
  
Wait for I/O to fill the next record before giving up on being able to
check  for repeater records.  Then double-check the records, if either
the repeater-flag is lit or if  the  record  sequence  number  of  two
consecutive  records  is  the same, treat them as repeater records and
ignore the first record.
********************************************************************************
  
  
                             EDIT 466    FOR DIRECT
  
[SYMPTOM]
  
MVOLUME may fail for interchange-format tapes.
  
  
[DIAGNOSIS]
  
Compare backwards.
  
  
[CURE]
  
Change 'CAIE' to 'CAIN'.
********************************************************************************
  
  
                             EDIT 107    FOR FILCOM
  
[SYMPTOM]
  
	Illegal memory reference with the /U switch.
  
[DIAGNOSIS]
  
	In some circumstances it is possible to back past useful
data and use random data as pointers.
  
[CURE]
  
	Add some checks so this does not happen.
  
  
********************************************************************************
  
  
                             EDIT 110    FOR FILCOM
  
[SYMPTOM]
  
	With /U lines get marked as update even though they
were not changed.
  
[DIAGNOSIS]
  
	Attempts to ignore blank lines were unsuccessful.
[CURE]
  
	Ignore the blank lines.  Also, print out an 'o' wherever
there have been deletions.
  
  
********************************************************************************
  
  
                             EDIT 111    FOR FILCOM
  
[SYMPTOM]
  
[DIAGNOSIS]
  
[CURE]
********************************************************************************
  
  
                             EDIT 236    FOR INITIA
  
[SYMPTOM]
  
	No way to set an account string fro INITIA created jobs.
[DIAGNOSIS]
  
	No code to do so.
[CURE]
  
	Add the "ACCOUNT" option to TTY.INI which specifies the account
string for an INITIA created job.  This option has no effect and is
specified for a terminal which will not be logged in by INITIA.  Additionally,
allot this to be specified on the "STOMP" command, and add a "FRCLIN"
********************************************************************************
  
  
                             EDIT 54     FOR PFH
  
[SYMPTOM]
  
  
     HALT in PFH at ONDSK-2.
  
[DIAGNOSIS]
  
     Range check from edit 53 is off by one.
  
[CURE]
  
     CAML becomes CAMLE.
********************************************************************************
  
  
                             EDIT 55     FOR PFH
  
[SYMPTOM]
  
	Another off-by-one bug in edit 53.
  
[DIAGNOSIS]
  
	When calculating the number of remaining words in a page.
  
[CURE]
  
  
	Edit 55 in PFH.
********************************************************************************
  
  
                             EDIT 326    FOR SYSTAT
  
[SYMPTOM]
  
	NO way to find out system uptime from SYSTAT on SMP system.
  
[DIAGNOSIS]
  
	SYSTAT only reports CPU uptimes which could be less than system
uptime if CPU's have been removed
  
[CURE]
  
	Have SYSTAT report system uptime as well as CPU uptimes on a SMP
system.
********************************************************************************
  
  
  
END OF  UTILITIES-10-V702