Google
 

Trailing-Edge - PDP-10 Archives - bb-bt99e-bb - utl702.d08
There is 1 other file named utl702.d08 in the archive. Click here to see a list.
                 EDIT DESCRIPTIONS FOR UTILITIES-10-V702
  
  
                             EDIT 500    FOR BACKUP
  
[SYMPTOM]
  
None.
  
  
[DIAGNOSIS]
  
Set up for version 4A
  
[CURE]
  
None.	
********************************************************************************
  
  
                             EDIT 501    FOR BACKUP
  
[SYMPTOM]
  
BACKUP will not restore files to SFDs if the upper level SFDs  or  UFD
contains  a file of the same name and is the same age or newer and the
user has /SCAN set via SETSRC.
  
  
[DIAGNOSIS]
  
Using the /SCAN switch to determine supersede criteria during restores
is incorrect.
  
  
[CURE]
  
In module BACKRS, in routine CHKSUP, set  the  .PTSCN  switch  in  the
.PTSWT word of the PATH block used by LOOKUP, while checking to see if
a file should be superseded.  This  prevents  the  scanning  of  upper
level  SFDs  or  the UFD.
********************************************************************************
  
  
                             EDIT 502    FOR BACKUP
  
[SYMPTOM]
  
If BACKUP  is  protected  execute-only,  a  proprietary  violation  is
reported when a WHAT command is typed after a SAVE is done.
  
  
[DIAGNOSIS]
  
Before calling the BACKRS module to do the actual SAVE, BACKUP does  a
CORE  UUO to get rid of the high segment to reduce the job size.  When
the SAVE is done and a WHAT command is typed, SCAN's routine KEYWDJ is
called  to scan for arguments.  Since it is in the high segment (which
isn't there) and the JS.XO bit is on, and there is no  PORTAL  in  the
entry  point  in  SCAN  (which are many), the monitor gives the user a
proprietary violation.
  
  
[CURE]
  
If BACKUP is execute-only, do not throw away the high  segment.   This
will  solve  all  transportation  problems  between  the  low and high
segment, especially to SCAN high segment code.
********************************************************************************
  
  
                             EDIT 503    FOR BACKUP
  
[SYMPTOM]
  
BACKUP will not restore files which were saved from ERSATZ devices.
  
  
[DIAGNOSIS]
  
The DEVNAM UUO does not return the physical device name if  passed  an
ERSATZ  device  and  that  device  is not in the job search list.  The
DSKCHR UUO will return the structure name given an ERSATZ device name.
  
  
[CURE]
  
In module BACKRS, in routine SETSTR, replace the DEVNAM UUO  with  the
DSKCHR  UUO  and  appropriate  arguments.  Also, skip the logical name
translation if "ALL" is the structure name.
********************************************************************************
  
  
                             EDIT 504    FOR BACKUP
  
[SYMPTOM]
  
Edit 373 was too restrictive of operator responses to a batch job.
  
  
[DIAGNOSIS]
  
Normally the operator would only want to give the answer GO, but
he might also want to say KILL in case he wants to deny the job an
extra reel of tape.
  
  
[CURE]
  
Edit 373 took too many features away.  Restore the check for batch
jobs before reporting error messages in order to precede them with
a $ sign.  Restore allowing the operator to type any runtime command.
Solve the original problem (see routine OPRCHS) of allowing numeric
characters to cause breaks.  Only comment characters, spaces, tabs and
real break characters will stop parsing.  Delete the E$$ICG error, since
it was redundant.
********************************************************************************
  
  
                             EDIT 505    FOR BACKUP
  
[SYMPTOM]
  
Minor version numbers greater than 26 ("Z") are output to the  listing
file as garbage.
  
  
[DIAGNOSIS]
  
BACKUP assumed that minor version numbers could only be one alphabetic
character.
  
  
[CURE]
  
Add code in LSTVER so that minor version numbers greater than  26  are
supported.
********************************************************************************
  
  
                             EDIT 506    FOR BACKUP
  
[SYMPTOM]
  
When writing a tape to be used at another site, it is a good  idea  to
make sure the tape is error free.
  
  
[DIAGNOSIS]
  
BACKUP defines the maximum number of tape  errors  to  be  100  before
giving  a  fatal  error  (a  warning  is  given every time an error is
detected).  To change the maximum requires BACKUP  to  be  edited  and
rebuilt.
  
  
[CURE]
  
Implement a new switch, ERRMAX n,  where  n  is  a  required  positive
decimal  number.   If  the  switch  is  not  given  explicitly  or  in
SWITCH.INI, then the old setting of 100 will be the default.
********************************************************************************
  
  
                             EDIT 507    FOR BACKUP
  
[SYMPTOM]
  
None.
  
  
[DIAGNOSIS]
  
Code looks wrong.  The definition of NM$TBF is documented in BACKRS to
agree  with  the  NM$TBF definition in BACKUP.  The definitions do not
agree.
  
  
[CURE]
  
Make the definitions agree.  Change the definition of NM$TBF in BACKUP
to agree with BACKRS.
********************************************************************************
  
  
                             EDIT 510    FOR BACKUP
  
[SYMPTOM]
  
"?Illegal data mode for device" error message is given by the  monitor
when restoring files from a tape created on another TOPS10 system.
  
  
[DIAGNOSIS]
  
The other system had defined their own data modes.  Files  written  in
these data modes had been saved on the tape.  BACKUP got the data mode
of the file being restored and indiscriminately did a SETSTS with  the
bit  pattern  found  on  the  tape.   Since  these data modes were not
defined in the running monitor, the job got an "?Illegal data mode for
device" error.
  
  
[CURE]
  
Before the SETSTS, do a DEVCHR of the  channel  OPENed  for  the  file
being  restored.  If the data mode stored on the tape is not legal for
the device the file is  being  restored,  give  the  warning  "%BKPIDM
Illegal  data  mode nn for file xxxx, assuming image mode" where nn is
the data mode stored on the tape and  xxxx  is  the  full  file  spec.
Restore the file setting the data mode to image.
********************************************************************************
  
  
                             EDIT 511    FOR BACKUP
  
[SYMPTOM]
  
Customers/DEC adding new switches to BACKUP can get confused  if  they
define  the  storage  following location F$MSN.  If the switch calls a
SCAN routine (.SWDEC, for example) as part of an SP macro,  SCAN  will
not  only  store  the  value  at  the storage location given in the SP
macro, but also at 2 words prior to that location.
  
  
[DIAGNOSIS]
  
BACKUP defines the "P$xxx" and "F$xxx" words in the wrong order.  SCAN
expects  "P$xxx"  to  follow "F$xxx";  in BACKUP this is not the case.
This causes SCAN to believe that the new switch storage is part of the
per-file  area (checked at FILSW1 in SCAN).  SCAN computes SWTPFO (the
offset to the sticky per-file-switches) by subtracting  the  addresses
of  F$xxx  from  P$xxx.  This causes the value of the new switch to be
stored at location minus the length of the per-file areas (in BACKUP's
case, 2).
  
  
[CURE]
  
Move the definitions  of  P$MBF  and  P$MSN  after  the  F$xxx.   Also
document this aspect of SCAN in BACKUP's comments.
********************************************************************************
  
  
                             EDIT 512    FOR BACKUP
  
[SYMPTOM]
  
Users cannot find a particular  version  of  a  file  when  using  the
listing features of BACKUP.
  
  
[DIAGNOSIS]
  
BACKUP does not list version numbers in the list files.
  
  
[CURE]
  
Change BACKUP to list the version number of  a  file  if  the  version
exists.   The  version  number  appears  after the date and before the
structure name as in the following example:
  
Start of save set  on MTA260 SCHED
System     RC177B KL #1026/1042 TOPS-10 monitor 702(24353) APR#1026
800 BPI 9 track  4-Nov-82 13:57:05 BACKUP 4A(512) tape format 1
Tape number  1
  
  
 BLIS10 EXE     356     <155>   16-Feb-78       7E(227) NEW:    [1,5]
********************************************************************************
  
  
                             EDIT 513    FOR BACKUP
  
[SYMPTOM]
  
BACKUP's PRINT command occasionally lists  file  lengths  incorrectly.
(The KL-10 V231 Microcode distribution tape is a good example.)
  
  
[DIAGNOSIS]
  
An ASCII file is saved (by DUMPER), storing a byte size of seven,  the
actual  file  length in words, and a data mode of image.  When a PRINT
command is typed, BACKUP calculates the size of the file by  examining
the  data  mode of the file.  If the data mode is ASCII (mode 0 or 1),
BACKUP then divides the size of the file (stored on the tape in words)
by  five.   DUMPER had stored the data mode of some of the ASCII files
on the KL-10 V231 Microcode distribution tape as 10 (image mode).  The
file  size for these files was computed incorrectly because BACKUP did
not divide the number of words by 5.
  
  
[CURE]
  
Change the alogrithm of computing the file size to  be  based  on  the
byte  size  of  the file instead of ASCII data modes.  The formula has
been taken from DIRECT which does the computation correctly.
********************************************************************************
  
  
                             EDIT 514    FOR BACKUP
  
[SYMPTOM]
  
  
A SKIP n command, where n is greater than the number of tape marks  on
the tape, causes the tape to go off the reel.
  
  
[DIAGNOSIS]
  
Physical EOT is only  seen  on  write  operations.   Since  a  magtape
skipfile UUO is a monitor read operation, the tape does not stop until
a tape mark is encountered for every logical  SKIP  command.   If  the
tape  has  been  positioned  after the last tape mark, another magtape
skipfile UUO causes the tape to spin until it goes off the reel.
  
  
[CURE]
  
If the argument to the SKIP command is zero or  negative,  do  not  do
anything special.  If, and only if, the label type is USER-EOT and the
command argument is positive, do a skip record before  the  skip  file
and  check  for  a tape mark.  If there is more data on the tape, then
the skip file operation can be executed.  If EOF  gets  flagged,  then
the  tape is now positioned AFTER the logical EOT.  If EOF is reached,
reposition the tape BEFORE the last tape mark, tell the user
  
     %BKPAMP Attempt to move past logical EOT
  
and continue processing the SKIP command in case tape  arguments  were
used.
  
The error handling routine in the area of this code has been  reworked
to ensure errors are actually detected for the specific skip operation
which caused them.
********************************************************************************
  
  
  
  
                             EDIT 2065   FOR LOGOUT
  
[SYMPTOM]
  
	THE STATISTICS TYPED BY LOGOUT DO NOT MATCH THOSE
	PRODUCED BY THE ACTDAE
  
[DIAGNOSIS]
  
	CHARGE FOR EVERTHING NOT JUST BATCH
  
[CURE]
  
		MAKE THE STATISTICS TYPED BY LOGOUT MATCH THOSE
		PRODUCED BYT THE ACTDAE
********************************************************************************
  
  
  
  
                             EDIT 50     FOR PFH
  
[SYMPTOM]
  
  
Halt or ill mem ref from PFH when doing dump mode  I/O  with
an  indexed  or  indirected address in a IN or OUT UUO while
virtual.
  
  
                        **********
[DIAGNOSIS]
  
We fake up a UUO to execute and supply a pointer to the IOWD
we  like,  but  forget  to clear any indirecting or indexing
bits in the original UUO.
  
  
                        **********
[CURE]
  
When fetching the UUO from the user program, clear the index
or indirect bits.
********************************************************************************
  
  
                             EDIT 51     FOR PFH
  
[SYMPTOM]
  
  
[DIAGNOSIS]
  
  
[CURE]
  
    Keep DATEND pointing into page 776. If it overflows a little, it
    doesn't matter since first few words of 777 aren't used.
********************************************************************************
  
  
                             EDIT 52     FOR PFH
  
[SYMPTOM]
  
	Fix up GETWRD some more, allow for the uuo itself to be in AC 1.
	Don't use EA as an indirect, use only as a pointer. If the word
	pointed to by EA is desired, it is available in AT.EA, filled in
	by GETWRD.
  
[DIAGNOSIS]
  
  
[CURE]
********************************************************************************
  
  
                             EDIT 53     FOR PFH
  
[SYMPTOM]
  
  
When running a program that does dump mode I/O virtual:
  
1) ?Halt at user PC 777472
  
2) ?Illegal instruction at user PC 777426
  
  
                        **********
[DIAGNOSIS]
  
1) Dump mode IOWDs are not broken down correctly if they end
in the last data page of the program.
  
2) Dump mode IOWDs are not correctly traced down if the path
falls  through  ACs 1-4, or involves indirection or indexing
with these ACs.
  
  
                        **********
[CURE]
  
1) Add some code  around  TSTDM7  to  notice  when  to  stop
checking for pages being accessible.
  
2) Rework a lot of code that  calls  GETWRD  and  references
variable EA, to avoid doing any address calculations outside
the routine GETWRD.
********************************************************************************
  
  
  
  
                             EDIT 176    FOR REACT
  
[SYMPTOM]
  
     Edit 165 causes incorrect  entry  to  be  used  as  source  in  I
PPNDST=PPNSRC command if PPNSRC>PPNDST.
  
  
[DIAGNOSIS]
  
     PPTAB offset of PPNSRC is found before  PPNDST  is  inserted  (if
necessary).   This  makes  the  old offset of PPNSRC invalid after the
call to INSSUB if PPNSRC>PPNDST (ie PPNSRC lies above PPNDST in  PPTAB
and hence is displayed by INSSUB).
  
  
[CURE]
  
     Check and adjust the offset PPNSRC if the tasks are expanded.
  
  
********************************************************************************
  
  
                             EDIT 177    FOR REACT
  
[SYMPTOM]
  
     ENQ/DEQ quotas specified by explicit numbers are read and  stored
correctly.   For  example,  explicitly specifying and ENQ/DEQ quota of
511 results in the number 23.
  
  
[DIAGNOSIS]
  
     EDIT 162 is incomplete.  If <CR> is typed, the default  is  used,
but if anything else is typed, the default is not cleared.
  
  
[CURE]
  
     Check to see if there is a default, if so clear it.
  
  
********************************************************************************

  
				EDIT 36 FOR UFDSET
  
  
  [SYMPTOM]
 
 
     Undeserved RENAME error  26  messages  when  trying  to  mount  a
structure with NOWRITE status.
 
 
                        **********
[DIAGNOSIS]
 
     UFDSET sets the structure status to NOWRITE before attempting the
rename.
 
 
                        **********
[CURE]
 
     Set the status after the rename.
 
  
********************************************************************************
  
  
  
END OF  UTILITIES-10-V702