Google
 

Trailing-Edge - PDP-10 Archives - BB-4172G-BM - 4-documentation/pat.tco
There are 15 other files named pat.tco in the archive. Click here to see a list.
TOPS20 Change Order Number 4.1887

Written by: DBELL		28-Feb-78 16:26:28

Edit checked: NO			   Document: NO
  TCO Tested: NO		Maintenance Release: NO
Hardware-related: NO

Program: PAT

Related TCO's:
Related SPR's: 


Problem: FILES CAN'T BE FORCED OUT TO DISK

Diagnosis: NO CODE

Solution: IMPLEMENT FILOP. FUNCTION 10.  THIS WILL UPDATE ALL
PAGES TO THE DISK, AND SET THE END OF FILE POINTER.
TOPS20 Change Order Number 4.1898

Written by: HURLEY		 7-Apr-78 17:01:11

Edit checked: NO			   Document: NO
  TCO Tested: NO		Maintenance Release: NO
Hardware-related: NO

Program: PAT

Related TCO's:
Related SPR's: 


Problem: REWINDS DID NOT WORK ALWAYS

Diagnosis: BOT BIT WAS NEVER CLEARED AT GSTATS

Solution: CLEAR BOT BEFORE UPDATING THE STATUS AT GSTATS
TOPS20 Change Order Number 4.1899

Written by: HURLEY		11-Apr-78 14:10:18

Edit checked: NO			   Document: NO
  TCO Tested: NO		Maintenance Release: NO
Hardware-related: NO

Program: PAT

Related TCO's:
Related SPR's: 


Problem: MAG TAPE BUFFER SIZE IS SET UP WRONG
	FOR ANSI-ASCII MODE.

Diagnosis: PAT READS THE USER'S DATA MODE AND THEN GETS THE DEFAULT
	RECORD LENGTH.  USING THESE VALUES IT CALCULATES THE NUMBER OF
	WORDS PER BUFFER.  WHAT IT SHOULD DO IS TO USE THE DEFAULT
	VALUES OF BOTH THE MADE AND THE RECORD SIZE TO GET WORDS PER
	BUFFER.

Solution: 
TOPS20 Change Order Number 4.1987

Written by: KONEN		18-Aug-78 08:05:57

Edit checked: NO			   Document: NO
  TCO Tested: NO		Maintenance Release: NO
Hardware-related: NO

Program: PAT

Related TCO's:
Related SPR's: 


Problem: FILES SOMETIMES GET LOST WHEN GOING TO THE PRINTER

Diagnosis: FILES GOING DIRECTLY TO LPT: HAVE A ZERO BYTE COUNT, EVEN THOUGH
THERE IS AT LEAST ONE PAGE.  FILES GOING TO LPT: ARE COPIED BY SOUT'S, AND
DO NOT NEED TO HAVE THE BYTE COUNT CORRECTED WHEN CLOSING THE FILE.

Solution: AT UNMCM1: CHECK IF FILE IS DISK, SKIP TO NEXT JFN IF NOT.
TOPS20 Change Order Number 4.2014

Written by: HELLIWELL		19-Sep-78 10:55:53

Edit checked: NO			   Document: NO
  TCO Tested: NO		Maintenance Release: NO
Hardware-related: NO

Program: PAT

Related TCO's:
Related SPR's: 


Problem: PA1050 DOES NOT SUPPORT GETTABS .GTRDV(135) AND .GTRDI(136)
THESE RETURN THE DEVICE AND PPN FROM WHICH PROGRAM CAME.

Diagnosis: NOT NEEDED TILL NOW

Solution: GENERALIZE ROUTINE WHICH DETERMINES HIGH SEGMENT DEVICE
AND PPN TO WORK FOR LOW SEGMENT. ADD CODE FOR GETTABS.
TOPS20 Change Order Number 4.2017

Written by: HELLIWELL		20-Sep-78 17:36:19

Edit checked: NO			   Document: NO
  TCO Tested: NO		Maintenance Release: NO
Hardware-related: NO

Program: PAT

Related TCO's:
Related SPR's: 


Problem: PA1050 GETS ILLEGAL MEMORY WRITE IF HIGH SEGMENT IS WRITE PROTECTED

Diagnosis: PA1050 COPIES SOME INFO FROM LOW SEG JOB DATA ARE TO
HIGH SEG JOB DATA AREA ON EXIT UUO.

Solution: CHECK IF VESTIGIAL JOB DATA ARE IS WRITABLE BEFORE COPYING
DATA FROM LOW SEG. IF NOT MERELY SKIP IT.
TOPS20 Change Order Number 4.2018

Written by: HELLIWELL		21-Sep-78 10:25:37

Edit checked: NO			   Document: NO
  TCO Tested: NO		Maintenance Release: NO
Hardware-related: NO

Program: PAT

Related TCO's:
Related SPR's: 


Problem: GETTAB TABLE 5 NOT IMPLEMENTED

Diagnosis: ACTUAL VALUE DESIRED DOES NOT EXIST FOR TOPS20

Solution: IMPLEMENT TABLE AND RETURN A VALUE WHICH IS MONOTONICALLY
INCREASING WITH TIME. (I.E. RUNTIME * CONSTANT).
TOPS20 Change Order Number 4.2030

Written by: HURLEY		 3-Oct-78 11:15:57

Edit checked: NO			   Document: NO
  TCO Tested: NO		Maintenance Release: NO
Hardware-related: NO

Program: PAT

Related TCO's:
Related SPR's: 


Problem: LINE FEED ECHOS INCORRECTLY IN APL

Diagnosis: CCOC BITS WRONG FOR LINE FEED

Solution: MAKE THE CCOC BITS BE 2 INSTEAD OF 3 FOR LINE FEED
TOPS20 Change Order Number 4.2053

Written by: ALUSIC		19-Oct-78 10:38:53

Edit checked: NO			   Document: NO
  TCO Tested: NO		Maintenance Release: NO
Hardware-related: NO

Program: PAT

Related TCO's:
Related SPR's: 


Problem: SLEEP SLEEPS FOR THE INCORRECT LENGTH OF TIME, OFTEN OFF BY UP TO
10%.

Diagnosis: THIBR JSYS USED WHEN THE ARG IS IN SECONDS.  THIBR DOESNT WORK
CORRECTLY. NOTE THAT DISMS WORKS AND IS USED WHEN THE ARG IS LESS THAN 1 SEC.

Solution: REMOVE THE THIBR CODE AT IOWATA AND USE DISMS ONLY.
TOPS20 Change Order Number 4.2057

Written by: HELLIWELL		20-Oct-78 18:35:40

Edit checked: NO			   Document: NO
  TCO Tested: NO		Maintenance Release: NO
Hardware-related: NO

Program: PAT

Related TCO's:
Related SPR's: 


Problem: GETTAB TO GET HIGH SEGMENT ORIGIN FAILS

Diagnosis: GETTAB TABLE # 100 NOT IMPLEMENTED

Solution: IMPLEMENT GETTAB TABLE # 100 FOR CURRENT HIGH SEGMENT ONLY.
TOPS20 Change Order Number 4.2073

Written by: HELLIWELL		29-Oct-78 12:28:58

Edit checked: NO			   Document: NO
  TCO Tested: NO		Maintenance Release: NO
Hardware-related: NO

Program: PAT

Related TCO's:
Related SPR's: 


Problem: HI SEGMENT WITH A LENGTH WHICH IS EXACT MULTIPLE OF PAGES
GETS 1 PAGE LONGER ON A GETSEG.

Diagnosis: CODE AT RUN24+12 ROUNDS UP TO PAGE BOUNDARY, BUT DOESN'T
ACCOUNT FOR LENGTH BEING EXACT MULTIPLE OF PAGES.

Solution: MANY PLACES NEED TO UNDERSTAND THAT THE HI SEG
FREE SPACE POINTER INDICATES THE FIRST UNUSED WORD, NOT THE
LAST USED WORD. THIS MEANS IT CAN POINT OFF THE END OF THE HI SEG.
FIXED CODE AT RUN24, HSOCHK, VESTIG, AND REMAP1.
TOPS20 Change Order Number 4.2080

Written by: ALUSIC		 6-Nov-78 11:02:57

Edit checked: NO			   Document: NO
  TCO Tested: NO		Maintenance Release: NO
Hardware-related: NO

Program: PAT

Related TCO's:
Related SPR's: 	 11500


Problem: TTCALL 6 (GETLCH) WILL HANG IF THE PROG ASKING FOR LINE CHARACTERISTICS
IS DETACHED.  THIS WAS DISCOVERED IN TRYING TO RUN APLSF DETACHED.

Diagnosis: TOPS-10 RETURNS 0 WHEN THE CONTROLLING TTY IS DETACHED, TOPS-20
JUST WAITS TILL IT IS ATTACHED TO RETURN TERMINAL CHARACTERISTICS.  MAKE
PA1050 RETURN 0 IF THE JOB IS DETACHED.

Solution: AT TTCL6 CHECK TO SEE IF THE JOB IS DETACHED AND THEN RETURN 0 TO
CALLER IMMEDIATELY, DON'T GO THROUGH DVCHR AND RFMOD STUFF.
TOPS20 Change Order Number 4.2090

Written by: ALUSIC		17-Nov-78 09:46:13

Edit checked: NO			   Document: NO
  TCO Tested: NO		Maintenance Release: NO
Hardware-related: NO

Program: PAT

Related TCO's:
Related SPR's: 	


Problem: FORTRAN BACKSPACE DOES NOT WORK PROPERLY FOR MAGTAPE, IT SEEMS TO GET
CONFUSED WHEN BACKSPACING OVER PHYSICAL RECORDS.  ALSO FORTRAN REREAD DOES NOT
WORK PROPERLY.

Diagnosis: PA1050 DID NOT SET THE USE BITS IN THE INPUT BUFFER HEADERS WHEN THEY
SHOULD HAVE BEEN.

Solution: 
SETIB3+3/POPJ P,	JRST SETIB4

SETIB2-1/POPJ P,	SETIB4:	MOVSI A,400000
				XCTUU <IORM A,@(CC)>
				POPJ P,

;SET THEM
TOPS20 Change Order Number 4.2091

Written by: HELLIWELL		17-Nov-78 17:17:34

Edit checked: NO			   Document: NO
  TCO Tested: NO		Maintenance Release: NO
Hardware-related: NO

Program: PAT

Related TCO's:
Related SPR's: 


Problem: UNDER SOME CONDITIONS PA1050 LOOKS ON "DSK:<SUBSYS>"
FOR FILES WHEN "SYS:" IS NOT DEFINED.

Diagnosis: CODE USES "DSK:<SUBSYS>"

Solution: USE "PS:<SUBSYS>"
TOPS20 Change Order Number 4.2121

Written by: ALUSIC		13-Dec-78 14:35:34

Edit checked: NO			   Document: NO
  TCO Tested: NO		Maintenance Release: NO
Hardware-related: NO

Program: PAT

Related TCO's:
Related SPR's: 	 11700


Problem: THE EOF POINTER IS NOT ALWAYS A CORRECT REFLECTION OF THE NUMBER OF
WORDS IN THE FILE.  THE CORRECT NUMBER IS ALWAYS IN THE FILE BUT THE FILE IS 
PADDED WITH ZEROS TO A MULTIPLE OF 128 WORDS.  THIS SCREWS UP (AT LEAST) FORTRAN
BINARY RANDOM ACCESS FILES.

Diagnosis: AT MOVBUF WHEN DOING SIN /SOUT TO A DSK FILE 200 WORDS IS ALWAYS
ASSUMED.

Solution: REPLACE THE V2 CODE AT MOVBUF WHICH CALCULATES THE CORRECT EOF AND
NUMBER OF WDS TO TRANSFER.
TOPS20 Change Order Number 4.2133

Written by: HELLIWELL		27-Dec-78 11:56:23

Edit checked: NO			   Document: NO
  TCO Tested: NO		Maintenance Release: NO
Hardware-related: NO

Program: PAT

Related TCO's:
Related SPR's: 


Problem: REMAP UUO SOMTIMES SCREWS UP HI SEG

Diagnosis: WHEN MOVING CODE UP DURING REMAP, LOOP COUNT IS NOT SET UP

Solution: SET UP LOOP COUNT WHEN COMPUTING HI SEG LENGTH AT REMAP1.
TOPS20 Change Order Number 4.2169

Written by: ALUSIC		23-Jan-79 14:22:17

Edit checked: NO			   Document: NO
  TCO Tested: NO		Maintenance Release: NO
Hardware-related: NO

Program: PAT

Related TCO's:
Related SPR's: 	 12037


Problem: 
	APLSF USERS COULD NOT SET TERMINAL WIDTHS CORRECTLY, AND THE TER
WIDTHS WOULD GET CHANGED APPARENTLY FOR NO REASON.

Diagnosis: 	APLSF DOES A .TONFC TRMOP TO CLEAR FREE CRLF. PA1050 SETS
TERMINAL WIDTH TO 0 TO SIMULATE THIS AND SUBSEQUENTLY CHOOSES TO HANDLE 0
INCORRECTLY.  WHEN RETURNING TO THE MONITOR THE CRLF IS RESET BY APLSF -
(.TONFC) AND PA1050 USED RFMODS AND STPARS ETC WHICH LIMITED WIDTH FIELD TO
7 BITS AND ALSO DID OTHER STUFF WRONG.  MAX WIDTH WAS 127 RATHER THAN 255
ETC.

Solution: USE MTOPR IN THE SET AND CLEAR .TONFC TRMOP FUNCTION AS WELL AS THE
SET AND READ .TOWID FUNCTION, NOW THE FOLLOWING IS TRUE:
	CLEAR CRLF GIVES USER TERMINAL WIDTH 0
	SET CRLF 
		IF CURRENT WIDTH NONZERO GIVES CURRENT WIDTH.
		IF CURRENT WIDTH ZERO,
			IF CRLF CLEARED WITH .TONFC, GIVES WIDTH AT THAT TIME
			IF CRLF NOT CLEARED WITH .TONFC GIVES 72.
	READ WIDTH GIVES CURRENT WIDTH IF NONZERO, IF ZERO GIVES 255.
	SET WIDTH UP TO 255.
TOPS20 Change Order Number 4.2182

Written by: ALUSIC		 2-Feb-79 13:13:56

Edit checked: NO			   Document: NO
  TCO Tested: NO		Maintenance Release: NO
Hardware-related: NO

Program: PAT

Related TCO's:
Related SPR's: 	 11746


Problem: 
IF A REAL UNRECOVERABLE DISK ERROR OCCURS DURING A DBMS DSK ACCESS, PA1050 TRAPS
THE ERROR, PRINTS A WARNING(%) MESSAGE AND CONTINUES.  THE DBMS CODE DOES NOT 
SEE THE ERROR SINCE IT IS A HYBRED PROG WHICH HAS ISSUED ITS OWN JSYS' AND THE 
DSK JFNS ARE UNKNOWN TO PA1050.  THE RESULT IS A DAMAGED DATABASE WHICH CANNOT 
BE RECOVERED WITH JOURNALLING SINCE THE ERROR IS NOT KNOWN IN TIME.

Diagnosis: 
PA1050 CHAN 11 TRAP ROUTINE CHECKS TO SEE IF THE JFN IS KNOWN TO IT. IF YES IT
DOES ALL THE CORRECT TOPS-10 THINGD, IF NO IT ISSUES A WARNING ,DISMISSES THE 
INTERRUPT AND CONTINUES.  IT SHOULD REALLY HALT SO USERS CAN RECOVER DAMAGED
DATA.

Solution: 
AT IOERR:  GIVE FATAL MESSAGE AND HALT IF THE BUFFER HEADERS ARE NONEXISTANT, 
JFN IS NOT KNOWN OR NOT IN PAT.
TOPS20 Change Order Number 4.2196

Written by: WHITLOCK		28-Feb-79 10:26:17

Edit checked: NO			   Document: NO
  TCO Tested: NO		Maintenance Release: NO
Hardware-related: NO

Program: PAT

Related TCO's:
Related SPR's: 


Problem: Control characters are displayed on terminal input and output
as <up-arrow letter>.  TOPS-10 echos control characters on input with
<up-arrow letter> but allows control characters on output to be sent to 
the terminal unmolested.  APL and other languages need this behavior to
control graphics terminals.

Diagnosis: PA1050 sets the MODE word for the controlling terminal to
ASCII mode which means both input and output are displayed via the
CCOC words.

Solution: Edit 320 sets the MODE word to ASCII with no output translation
(.TTATE).  Define after TM.ASC:
	TM.ATE==3B29
and at TTYST0+5 lines, change:
	  TRZA B,TM.ASC
	TRO B,TM.ASC
to be:
	  TRZA B,TM.ATE
	TRO B,TM.ATE
TOPS20 Change Order Number 4.2197

Written by: WHITLOCK		28-Feb-79 10:33:09

Edit checked: NO			   Document: NO
  TCO Tested: NO		Maintenance Release: NO
Hardware-related: NO

Program: PAT

Related TCO's:
Related SPR's: 


Problem: PA1050 resets some fields in the MODE word before exiting on
^C but doesn't restore the user's MODE word on CONTinue.

Diagnosis: PA1050 resets TT%WAK, TT%ECO and TT%DAM before exiting on ^C.
On CONTinue, it does not reset these to the user's settings.

Solution: Edit 321 defines a storage word before the item DEVNM7:
		ALC SVMOD,1
	inserts at EXIT2+8 lines:
		MOVEM B,SVMOD
	and inserts at EXIT1+4 lines:
		MOVE B,SVMOD
		MOVEI A,PRIJFN
		SFMOD
TOPS20 Change Order Number 4.2198

Written by: WHITLOCK		28-Feb-79 10:39:39

Edit checked: NO			   Document: NO
  TCO Tested: NO		Maintenance Release: NO
Hardware-related: NO

Program: PAT

Related TCO's:
Related SPR's: 


Problem: OUTSTR TTCALL is very slow.

Diagnosis: PA1050 does a BOUT loop on the ASCIZ string to simulate the OUTSTR.
The BOUT code checks for TERMINAL INDICATE and sends ^L if necessary.

Solution: Edit 322 replaces the code to do a BOUT loop with code to set the
CCOC word for TERMINAL INDICATE and do a SOUT.  Replace the following 5
lines at TTCL3+6 lines
	HRLI D,440700
	ILDB B,D
	JUMPE B,CPOPJ
	PUSHJ P,TTYBOU
	JRST .-3
with the following code:
	RFCOC
	PUSH P,B
	PUSH P,C
	TRZ C,3B19
	TRO C,2B19
	TRZ B,3B25
	TLNN PF,L.INDF
	TROA B,2B25
	TRO B,1B25
	SFCOC
	HRRO B,D
	SETZ C,
	SOUT
	POP P,C
	POP P,B
	SFCOC
	JRST CPOPJ
TOPS20 Change Order Number 4.2199

Written by: WHITLOCK		28-Feb-79 10:45:05

Edit checked: NO			   Document: NO
  TCO Tested: NO		Maintenance Release: NO
Hardware-related: NO

Program: PAT

Related TCO's:
Related SPR's: 


Problem: APL users are used to hitting the BREAK or ATTN key to signal ^C.
This is available on a 2741 terminal, which TOPS-20 doesn't support.

Diagnosis: The BREAK or ATTN key sends a NULL character, for which PA1050
is not enabled for interrupt.  PA1050 is set to interrupt on ^C, only if
the user is trapping ^C.

Solution: Edit 323 makes the BREAK or ATTN key act as 2 ^C and immediately
gives control to the user program if trapping ^C.  The BREAK key will do
nothing if the program is not trapping ^C.  At SETPSI+9 lines, replace
	SKIPE CCIENB
	  ATI
with
	SKIPN CCIENB
	  JRST EDONE
	ATI
	HRLI A,.TICBK
	ATI
EDONE:
TOPS20 Change Order Number 4.2200

Written by: WHITLOCK		28-Feb-79 10:52:16

Edit checked: NO			   Document: NO
  TCO Tested: NO		Maintenance Release: NO
Hardware-related: NO

Program: PAT

Related TCO's:
Related SPR's: 


Problem: Edit 320 disabled translation of control characters on output.
Edit 322 used a SOUT for output instead of a BOUT loop.  After edit 322,
formfeed always comes out formfeed even if TERMINAL INDICATE is set.

Diagnosis: A SOUT with a MODE of .TTATE (no output translation) ignores
TERMINAL INDICATE and always sends a formfeed as a formfeed.

Solution: Edit 324 fixes up edit 322 so that the line to be output is
pre-scanned.  If a formfeed is found, the old BOUT loop is used (which
will check for TERMINAL INDICATE correctly).  If no formfeed is found, the
faster SOUT code is executed.  At TTCL#+6 lines, before edit 322, insert:
	MOVE C,D
	HRLI C,440700
	ILDB B,C
	JUMPE B,TTCL3A
	CAIE B,C.FF
	  JRST .-3
	HRLI D,440700
	ILDB B,D
	JUMPE B,CPOPJ
	PUSHJ P,TTYBOU
	JRST .-3
TTCL3A:
TOPS20 Change Order Number 4.2201

Written by: MILLER		 2-Mar-79 16:25:50

Edit checked: NO			   Document: NO
  TCO Tested: NO		Maintenance Release: NO
Hardware-related: NO

Program: PAT

Related TCO's:
Related SPR's: 


Problem: Code added in release 2 to "fix" monitor problem
is incompatible with labeled tapes.

Diagnosis: As above. In order to remove it the monitor
bug must first be fixed.

Solution: Fix monitor bug in MAGTAP and remove the code
around MTAPE0.
TOPS20 Change Order Number 4.2239

Written by: WHITLOCK		23-Apr-79 11:39:57

Edit checked: NO			   Document: NO
  TCO Tested: NO		Maintenance Release: NO
Hardware-related: NO

Program: PAT

Related TCO's:
Related SPR's: 


Problem: THE MTCHR. UUO RETURNS 0 IN THE LEFT HALF OF THE AC INSTEAD OF THE
LENGTH OF THE RECORD LAST READ.

Diagnosis: 
PA1050 WAS NEVER TAUGHT TO DO THIS.

Solution: 
GET THE LENGTH OF THE LAST RECORD ON BYTES AND DIVIDE BY THE BYTES PER
WORD BASED ON DATA MODE SINCE MTCHR. EXPECTS THE RECORD LENGTH IN WORDS.

EDIT 326 PUTS THE FOLLOWING CODE IN FROM MTCHR3-2 TO MTCHR1:

	PUSH P,C		;(326) SAVE DENSITY
	GDSTS			;(326) GET DEVICE STATUS
MTCHR4:				;(326)
	HLRZS C			;(326) RIGHT-JUSTIFY BYTE COUNT
	PUSH P,C		;(326) SAVE BYTE COUNT
	MOVEI B,.MORDM		;(326) READ TAPE DATA MODE (JFN ALREADY SET)
	MTOPR			;(326) SHOULD NEVER RETURN 0 == SYSTEM DEFAULT
	  ERJMP RETZER		;(326) SHOULDN'T HAPPEN
	HLRZ B,TPRDMT(C)	;(326) GET BYTES PER WORD
	POP P,C			;(326) GET BYTE COUNT
	IDIV C,B		;(326) CONVERT BYTES TO WORDS
	SKIPE D			;(326) ROUND UP TO NEXT HIGHER WORD
	  AOS C			;(326) IF NEEDED
	POP P,A			;(326) GET DENSITY
	HRL A,C			;(326) PUT IN WORD COUNT
	JRST STOTC1		;RETURN ANSWER
MTCHR3:	GDSTS			;(326) GET DEVICE CHARS (JFN ALREADY SET)
	LDB B,[POINT 2,B,28]	;(326) GET DENSITY
	SKIPN B			;(326) IS IT 0 ?
	  MOVEI B,3		;(326) YES, MAKE IT 3
	PUSH P,B		;(326) SAVE DENSITY
	JRST MTCHR4		;(326) C ALREADY SET BY GDSTS
MTCHR1:	MOVEI D,17		;SEARCH FOT THIS DEV INITED
TOPS20 Change Order Number 4.2285

Written by: WHITLOCK		13-Jun-79 13:34:00

Edit checked: NO			   Document: NO
  TCO Tested: NO		Maintenance Release: NO
Hardware-related: NO

Program: PAT

Related TCO's:
Related SPR's: 	


Problem: Under release 4 of PA1050, APLSF on an LA37 is not setting
the terminal characteristics to TERM NO LOEERCASE.

Diagnosis: TCO 4.2080 added code at TTCL6 to check if the terminal was
detached.  If it is, a GETLCH UUO (get line characteristics) will return
0.  The AC setup for the GETJI jsys set AC1 to 777777 instead of -1 and
got a JSYS error which was JFCLed.  The code then decides APLSF is detached
and returns 0 so APL does not reset the upper/lower case of the terminal.

Solution: At TTCL6 + 4, change the MOVEI A,-1 to SETO A,
TOPS20 Change Order Number 4.2290

Written by: YODER		15-Jun-79 13:34:41

Edit checked: NO			   Document: NO
  TCO Tested: NO		Maintenance Release: NO
Hardware-related: NO

Program: PAT

Related TCO's:
Related SPR's: 	


Problem: 
 A program which opens NUL: through the compatibility package and then does
an ENTER on that device with a zero filename gets a non-skip (error) return
when this should be valid, and a no-op, for the null device.

Diagnosis: 
 The routine DIRCHK looks at the bit HASDIR in table DEVTBL, which is on for
device NUL: (and must be for compatibility with the -10, as DEVCHR must work).
However what is actually desired from DIRCHK is to see if a filename is
required.

Solution: 
 Change DIRCHK so that it checks for device NUL: and gives a non-skip return.
TOPS20 Change Order Number 4.2350

Written by: YODER		26-Jul-79 17:07:16

Edit checked: NO			   Document: NO
  TCO Tested: NO		Maintenance Release: NO
Hardware-related: NO

Program: PAT
   Routines Affected:	UOPEN

Related TCO's:
Related SPR's: 


Problem: Edit 331 broke some normal operations with NUL:

Diagnosis: When doing OPEN or INIT to device NUL:, PA1050 thinks it needs
to wait until it has a filename before doing a GTJFN.  Later, at LOOKUP
time, it used to make a cancelling error (thinking NUL: was a directory
device) so everything worked.  Edit 331 fixed the second error but not
the first, so no GTJFN was ever being done for NUL:, which resulted in
I/O to unassigned channel errors.

Solution: Make routines UOPEN/UINIT do a GTJFN immediately.
TOPS20 Change Order Number 4.2435

Written by: EKLUND		 4-Sep-79 09:52:18

Edit checked: YES			   Document: NO
  TCO Tested: YES		Maintenance Release: NO
Hardware-related: NO

Program: PAT

Related TCO's:
Related SPR's: 	 13403


Problem: If one attempts to REWIND a spooled card reader file in FORTRAN,
then PA1050 may die with an illegal instruction trap (on a SFPTR).

Diagnosis: FOROTS causes a CLOSE to happen when the REWIND 2 is encountered.
If there happens to be a later generation number for the spooled file being
read, then the REWIND causes this file to be used next.  Otherwise, PA1050
gets quite unhappy since there is a file, but not the file it thinks it
wants.  The SFPTR JSYS fails unexpectedly (file does not exist), and there
is no ERJMP after it (an oversight).  When this problem is corrected,
another problem happens later when the first READ of data is attempted -
this is corrected by the patch as well.

Solution: Put in an ERJMP after the SFPTR to handle the error condition
with the REWIND, and later on set things up so that an appropriate error
return is taken (back to FOROTS) so that the FORTRAN program can trap the
condition successfully.  The error will be an end-of-file condition, which
is the same error which would be issued if there is no spooled file
availible in the first place.  In version 6 of FOROTS an attempt will be
made to handle this whole situation more reasonably, since no CLOSE will
be attempted on the REWIND when the compatibility package is eliminated.
This will cause the REWIND to do what one expects.
TOPS20 Change Order Number 4.2437

Written by: DBELL		 4-Sep-79 10:53:14

Edit checked: NO			   Document: NO
  TCO Tested: NO		Maintenance Release: NO
Hardware-related: NO

Program: PAT
   Routines Affected:	tapop

Related TCO's:
Related SPR's: 


Problem: 
The TAPOP function to read magtapes backwards does not work at all.

Diagnosis: 
PA1050 only has a magtape opened when it actually does IO.  But to
do a read backwards, a magtape drive must be open.  So the MTOPR that
is done always fails.  Even if it worked, the read-backwards state
would disappear when PA1050 released the tape again.

Solution: 
Instead of doing the MTOPR right away, remember in the flag word
whether or not we are to read backwards, and check the flag when
a read is done.  Also reset the flag when an OPEN is done.
TOPS20 Change Order Number 4.2451

Written by: EKLUND		10-Sep-79 16:25:38

Edit checked: NO			   Document: NO
  TCO Tested: NO		Maintenance Release: NO
Hardware-related: NO

Program: PAT

Related TCO's:
Related SPR's: 	 13330


Problem: If PA1050 gets control on a QUOTA EXCEEDED interrupt (FORTRAN,
COBOL, APL, etc.) a variety of strange things may happen, all due to the
fact that the compatibility package is left in an indeterminate state.
Some of the known possible problems are corrupted workspaces (APL),
CATASTROPE during compilation (COBOL), etc.

Diagnosis: The code in PA1050 is quite incorrect, saving off the ACs,
but not leaving the stack pointer in a state which is usable for the
code which goes off to expunge deleted files.  The result is that the stack
(and possible some registers) are wrong upon the resumption after the 
expunge is attempted.

Solution: Fix up the stack pointer so that the intended action is possible.
This is quite a simple matter of resetting register 17 after using it to
BLT the ACs to a safe place.
TOPS20 Change Order Number 4.2492

Written by: DBELL		27-Sep-79 13:46:01

Edit checked: NO			   Document: NO
  TCO Tested: NO		Maintenance Release: NO
Hardware-related: NO

Program: PAT
   Routines Affected:	MTCHR4

Related TCO's:
Related SPR's: 


Problem: 
The MTCHR UUO sometimes gives the error return with 0 in the ac.

Diagnosis: 
The MTOPR at MTCHR4+3 fails if the magtape is not opened, and the
ERJMP just goes to RETZER.

Solution: 
If the MTOPR fails, return what status we can.
TOPS20 Change Order Number 4.2566

Written by: DBELL		 7-Nov-79 10:49:45

Edit checked: NO			   Document: NO
  TCO Tested: NO		Maintenance Release: NO
Hardware-related: NO

Program: PAT
   Routines Affected:	filop

Related TCO's:
Related SPR's: 


Problem: 
FILOP. function 10 to "update RIBs" does not work.

Diagnosis: 
Code does an UFPGS to force all pages back to disk, but the count
specified was 1,,0 (the correct value).  However, the monitor does
a HRRZ to get the count, thus thinking a count of 0 was specified!!
So the UFPGS does nothing.

Solution: 
The monitor cannot be changed, since documentation (now) says the
count goes in the right half.  So change PA1050 to use 777777 as
the count, which misses the last page (which is ok, since no TOPS-10
file is that large).
TOPS20 Change Order Number 4.2568

Written by: DBELL		 8-Nov-79 15:12:54

Edit checked: NO			   Document: NO
  TCO Tested: NO		Maintenance Release: NO
Hardware-related: NO

Program: PAT
   Routines Affected:	INIBUF

Related TCO's:
Related SPR's: 


Problem: 
Under release 4, LIBOL fails to read EBCDIC labeled tapes correctly.
Junk gets returned in the records.

Diagnosis: 
PA1050 always did I/O using 36 bit bytes.  But for labeled tapes,
the monitor changes the byte size to 8 bit bytes.  Thus, the SINR
done at SEQMTR stores only one character per word in the buffer,
instead of the correct 4.

Solution: 
Have PA1050 do a RFBSZ in the INIBUF routine if this is a magtape,
and use the bytesize to set up the SINR pointer and byte counter.