Enter your search terms Submit search form
	Web pdp-10.trailing-edge.com

                 EDIT DESCRIPTIONS FOR LINK-10-V5-1                             
  
  
                             EDIT 2076   FOR LINK
  
[SYMPTOM]
  
A command line to LINK that contains a file specification followed  by
a switch that requires an output file then a /PLOT switch will produce
a ?LNKINS I/O DATA BLOCK NOT SET UP.
  
[DIAGNOSIS]
  
The code in %PLOT uses NULSPC  to determine if the file  specification
has already  been seen.   If  NULSPEC is  non  zero, then  it  wrongly
assumes that this is  not the first time  the routine has been  called
with this file specification.  Other switches on the same line such as
/MAP may previously have set NULSPC.  If NULSPC has been set  already,
%PLOT behaves as though the I/O data block has already been set up.
  
[CURE]
  
Change the test in %PLOT to  use IO.PTR+PC not NULSPC to determine  if
********************************************************************************
  
  
                             EDIT 2077   FOR LINK
  
[SYMPTOM]
  
The /NEWPAGE switch does not work correctly.  Neither the symbol table
nor the patch area  start on a  page boundary despite  the use of  the
/NEWPAGE switch.
  
[DIAGNOSIS]
  
The /NEWPAGE switch sets the value of RC.CV (the current value of  the
relocation counter  for  the  segment)  to  the  next  page  boundary.
However, the code to determine the  postion of the patch area and  the
symbol table checks RC.HL ( the highest location loaded).
  
[CURE]
  
Have the code in .NEWPAGE set both RC.CV and RC.HL to the beginning of
********************************************************************************
  
  
                             EDIT 2100   FOR LINK
  
[SYMPTOM]
  
No long error text appears for a LNKUSB error.
  
  
[DIAGNOSIS]
  
The error test is missing in LNKERR.
  
  
[CURE]
  
********************************************************************************
  
  
                             EDIT 2101   FOR LINK
  
  
  
  
[SYMPTOM]
  
A null program name will appear as a space in the PDV.
  
  
[DIAGNOSIS]
  
The conversion/deposit subroutine DVDPB. always converts and stores at
least  one character, even if the sixbit word being converted is null.
This is an off-by-one loop problem.
  
  
[CURE]
  
Put the loop test at the top of the loop in DVDPB..  This insures that
the  correct  number of loops occur, and will not write a space to the
PDV for a null program name.
********************************************************************************
  
  
                             EDIT 2102   FOR LINK
  
[SYMPTOM]
  
LINK will not initialize a character DATA element falling on the page
boundary every multiple of ten pages in the .REL file.
  
[DIAGNOSIS]
  
LINK uses a ten-page buffer to read .REL files.
  
If the byte string to be used in initializing memory is split across
buffers, LINK moves the first part of the byte string, gets the next
buffer, resets the source byte pointer, and then moves the remainder
of the string.  The source byte pointer is constructed by taking the
position and bytesize information from the .REL file and the address
part from the .REL file buffer header. The .REL file buffer header
itself contains a fullword byte pointer.  The address part of this
fullword byte pointer is incremented by one before being used in the
source byte pointer, which is an ILDB-type pointer (with octal 44
in the position field).
  
However, if the byte string starts exactly at the beginning of the
new buffer, the .REL buffer's byte pointer is also an
ILDB-type pointer and incrementing the address causes the first word
of the byte string to be missed.
  
[CURE]
  
If the .REL buffer's byte pointer is "unused" (has octal 44 in the
position field) do not increment the address part when setting up
the source pointer for static byte string initialization.
  
********************************************************************************
  
  
  
END OF  LINK-10-V5-1