Trailing-Edge
-
PDP-10 Archives
-
bb-d549g-sb
-
maklib.man
There is 1 other file named maklib.man in the archive. Click here to see a list.
MAKLIB User's Guide V1.0
1. Introduction
MAKLIB is a program used to manipulate relocatable binary files
which are either single programs or collections of programs
(libraries). The relocatable binary file is the machine readable
output from a source language translator, such as the FORTRAN
compiler or the MACRO assembler. MAKLIB is used to manipulate the
individual modules within a library, to examine individual modules
and libraries and to make changes to modules when they need
correction. MAKLIB can also be used to create libraries by
deleting unnecessary symbols and by adding an index, which speeds
up loading by LINK-10.
2. Command String Format
The MAKLIB program uses the traditional DECSYSTEM-10 command
string, supported by use of the common SCAN and WILD routines. In
its most general format, the command string is:
OUTFILE = INFILE1/Switch/Switch.., INFILE2/Switch/Switch..
All MAKLIB commands require at least the output file and the first
input file. This first input file is called the MASTER file. If
a second input file is used by a command, this file is called the
TRANSACTION file. Some command switches do not require a
TRANSACTION file, particularly those that query the status of the
MASTER file. The standard SCAN supplied switches and features are
supported, including indirect command files and the use of
SWITCH.INI to initialize the command string. All switches can be
abbreviated to uniqueness, and comments are included in the
command line be preceding them with ";" or "!".
3. Command Switches That Give Information About the MASTER File.
The information commands yield data on the status and
contents of the file specified as the MASTER file. No TRANSACTION
file is specified with these command switches and they take no
arguments. The output file is an ASCII file with the default
extension .LST. The MASTER file has the default extension .REL.
1. LIST - This command switch is used to list the names of the
modules that are in the MASTER file.
EXAMPLE:
OUT.LST=MASTER.REL/LIST
2. POINTS - This command switch lists all the ENTRY points in the
specified MASTER file. ENTRY points are used by LINK-10 to
determine if a GLOBAL request can be satisfied by loading a
module from a library. The ENTRY points are usually
Page 2
subroutine starting addresses.
EXAMPLE:
OUT.LST=MASTER.REL/POINTS
3. TRACE - This command lists all the information contained in
the TRACE blocks in the specified MASTER file. These blocks
(Link item type 1060) are created by MAKLIB when the FIX
command is used to insert edits into a module in the file.
The TRACE blocks contain information about which edits are
installed and exactly what changes were made to the file. By
use of the TRACE command, the exact binary patching status of
the MASTER file can be ascertained.
EXAMPLE:
OUT.LST=MASTER.REL/TRACE
4. Commands Used to Create Library Files.
These two switches are used primarily to create libraries of
subroutines which are loaded when a main routine requests them.
An example of such a library is FORLIB.REL which contains the
various utility routines used by programs compiled by the FORTRAN
compiler. If a program requires a routine for a specific
function, it makes a GLOBAL request indicating this. At load
time, LINK-10 searches a set of user-specified and/or default
libraries looking for routines to satisfy the GLOBAL requests. A
library can be simply a set of routines concatenated into a single
physical file, however the use of the following two commands
greatly speeds up the searching of libraries. The command
switches are used on the MASTER file specification and there is no
TRANSACTION file. Neither command takes an argument and both can
be specified in a single command string. The default extension
for both the OUTPUT and the MASTER file specification is .REL.
1. INDEX - This command reads the MASTER file and produces an
OUTPUT file which is identical to the MASTER file except that
INDEX blocks are inserted directly in front of the first
module in the file. The INDEX blocks (Link item type 14)
provide a fast way for LINK-10 to find out what ENTRY points
are contained in the file, and where they are in the file.
LINK can then make searches more quickly, without having to
read the entire file, and if a module that is needed is in the
file, LINK knows where it can be found. This speeds up
loading by reducing the amount of I/O necessary.
EXAMPLE:
OUT.REL=MASTER.REL/INDEX
Page 3
2. NOLOCALS - This command switch causes an output file to
be created, identical to the MASTER file except that all LOCAL
symbols are deleted from the SYMBOL blocks (Link item type 2).
These LOCAL symbols are used primarily for debugging purposes
and by the FIX command in MAKLIB. Since they serve no purpose
in a production-mode library, they are often deleted to reduce
the amount of mass storage space the library occupies. The
deletion of LOCAL symbols also speeds up loading because less
I/O has to be done. GLOBAL symbols are left in the symbol
table, since they are needed for inter-linking of modules.
EXAMPLE:
OUT.REL=MASTER.REL/NOLOCALS
NOTE
The old FUDGE2 program, which MAKLIB supercedes, deleted
LOCAL symbols whenever an INDEX was created. MAKLIB does
not do this. To produce an indexed library without LOCAL
symbols the two command switches are given together:
OUT.REL=MASTER.REL/INDEX/NOLOCALS
5. Commands to Manipulate Library Modules
To facilitate the handling and creation of program libraries,
MAKLIB includes command switches that manipulate individual
routines in a collection of routines. Most of these commands
require the specification of a TRANSACTION file , along with the
OUTPUT and the MASTER files. The default extension for all three
files is .REL. These command switches take arguments, which are
the names of the modules that are to be affected by the switch.
1. MASTER:(Mod1,Mod2..) - This switch causes no manipulation of
the file. Its purpose is to set up a correspondence between
modules in the MASTER file and those that are specified in
some command switch which is specified for the TRANSACTION
file. An example is the REPLACE switch. The REPLACE switch
is given as part of the TRANSACTION file specification and the
arguments to this switch are names of modules which are to
replace those specified as arguments to the MASTER switch in
the MASTER file specification. The MASTER switch is always
given as part of the MASTER file specification , takes at
least one argument and requires that another command be given
in the same command string.
EXAMPLE:
OUT.REL=MASTER.REL/MASTER:Mod1, TRANS.REL/REPLACE:Mod1
Page 4
2. APPEND:(Mod1,Mod2..) - This command switch is used to add
new routines to the end of an existing library. The OUTPUT
file is created by copying the entire MASTER file into it,
followed by the routines specified as arguments to the APPEND
switch. The appended routines are read from the TRANSACTION
file, and must be in the same physical order as specified to
the switch.
EXAMPLE:
OUT.REL = MASTER.REL, TRANS.REL/APPEND:(Mod1,Mod2)
3. DELETE:(Mod1,Mod2...) - This command switch does not require
a TRANSACTION file specification, and must be given as part of
the MASTER file specification. The action taken is that the
entire MASTER file is copied to the OUTPUT file except for the
routines named as arguments to the DELETE switch. The
arguments to the DELETE switch must be names of modules in the
MASTER file, and they must appear in the same order as the
physical order of the modules in the file.
EXAMPLE:
OUT.REL = MASTER.REL/DELETE:(MOD1,MOD2)
4. EXTRACT:(Mod1,Mod2...) - This command can be used to produce
an OUTPUT file containing a subset of the routines in the
MASTER file. The switch does not require that a TRANSACTION
file specification be given as part of the command string.
The modules specified as arguments to the EXTRACT switch are
copied out to the OUTPUT file. They are copied out in the
order specified, which must correspond to their physical order
in the MASTER file.
EXAMPLE:
OUT.REL = MASTER.REL/EXTRACT:(MOD1,MOD2)
5. INSERT:(Mod1,Mod2...) - This command is used to insert new
modules into the library. This command requires the use of
the MASTER switch. The OUTPUT file is formed in the following
manner. The MASTER file is copied to the OUTPUT file up until
the , but not including, the module named as the first
argument to the MASTER switch. Then the module named as the
first argument to the INSERT switch is copied from the
TRANSACTION file to the OUTPUT file. The process is then
repeated, until the argument list specified to the MASTER
switch is exhausted, at which point the remaining parts of the
MASTER file are copied to the OUTPUT file. There must of
course be one argument to the MASTER switch for every argument
to the INSERT, and vice versa. To insert more than one module
in front of the same module in the MASTER, the MASTER module
name must appear repeatedly in the argument list. The order
of the module names in the argument lists of both the MASTER
and the INSERT switches must correspond to the physical order
Page 5
of the modules in the MASTER and TRANSACTION files ,
respectively.
EXAMPLE:
OUT.REL = MAST.REL/MASTER:MOD1, TRANS.REL/INSERT:MODA
6. REPLACE:(Mod1,Mod2...) - This command is used to replace
modules in the MASTER file with those specified by the REPLACE
switch. The OUTPUT file then is comprised of the entire
MASTER file with some modules replaced by those read from the
TRANASCTION file. Use of the REPLACE switch requires the use
of the MASTER switch so that MAKLIB knows which master modules
are to be replaced with those specified in the REPLACE
command. There must be a one to one correspondence in the
number of MASTER arguments and REPLACE arguments. The order
of both argument lists must also correspond to the physical
order of the modules in the related files.
EXAMPLE:
OUT.REL=MAST.REL/MASTER:Mod1,TRANS.REL/REPLACE:Moda
6. Editing the MASTER file
1. FIX - This command indicates that the MASTER file is to be
changed.
The FIX command switch is used to make changes to the
actual code and symbol table of a routine or program. The FIX
command is given as part of the TRANSACTION file
specification, and the only other switch allowed is the WHO
switch. The default extension for both the OUTPUT and the
MASTER file is .REL while the default for the TRANSACTION file
extension is .FIX. The FIX switch does not take any
arguments, but instead indicates that the TRANSACTION file is
an ASCII file formatted to tell MAKLIB what changes to make to
what modules, and how to identify them.
The FIX file format is the same as that for ordinary
MACRO assembly language programs, with several restrictions
and with the addition of several special purpose pseudo-ops.
The special pseudo-ops are used to pass data to MAKLIB on
where and how to insert the code given in the FIX file. The
restrictions are that macros and repeats are not implemented,
along with some of the MACRO-10 pseudo-ops. In addition,
since the MAKLIB assembler is one pass, there are some
restrictions on the types of forward references allowed. A
description of each new pseudo-op follows.
Page 6
1. .EDIT nnnnnn
Gives the edit name for the fix that follows.
All the other MAKLIB pseudo-ops must appear
between an .EDIT and an .ENDE pseudo-op.
The edit name is stored in the TRACE
block for any module affected by the edit.
2. .NAME iii
Gives initials of person who wrote the fix (optional).
These initials, if present are stored in the
TRACE block for the edit.
3. .DATE dd-mon-yy
Gives the date that the fix was created (optional).
This date is remembered in the TRACE block for the edit.
4. .MODULE mmmmmm
Selects the module name that edit should be applied to.
The selected module is read in (loaded)
and remains the selected module until the end of
the edit, unless a new .MODULE pseudo-op is issued.
Within a given edit, each module may be selected
once at most.
5. .ASSOCIATED +edit1, -edit2, edit3, +edit4 .....
Conveys information about which other edits must be
present or not present in the current module.
The "+" indicates that the following edit is required.
The "-" indicates that the edit is precluded.
Warning messages are given if the module does not have
the appropriate combination of associated edits.
If no + or - is seen, + is assumed. (Edit required)
All .ASSOCIATED pseudo-ops must precede the first
occurence of any .INSERT,.REMOVE or .REINSERT operator.
6. .INSERT expression,KEYWORD:n,<Code to Match>
This pseudo-op tells where and how to insert code.
It also provides a method for verifying that the
position of the fix is correct.
The first argument gives the location at which to
install code, in either numeric or symbolic
expression, evaluated in the current radix.
This location is assumed to be relocatable and may
be in either the low or high segment. Any new code
inserted will be placed at the end of the same
segment.
The second argument is a keyword giving the method of
the installation. The keyword is one of the
following (abbreviated to uniqueness):
BEFORE - Execute patch code before executing
the instruction at the location specified.
AFTER - Execute patch code after executing the
instruction at the location specified.
REPLACE - For each instruction included in the
Page 7
patch, delete one from existing code, starting
with the instruction at the location of the
patch.
REPLACE:m - Delete m instructions from the existing
code starting with instruction at the patch
address, no matter how many instructions
are inserted. The argument may
be an expression, and is
evaluated in the current radix.
The third argument, which is optional, gives the line
of code at the location specified by the
first argument.
The expression within the angle brackets is evaluated,
and if it does not match the code actually at that
location, a fatal error message is given.
If the code being matched involves a multi-word
literal, only the first word need be given.
After the .INSERT pseudo-op, code generated by
the assembler is inserted, according to instructions.
7. .ENDI - This pseudo-op delimits the code that is
part of the preceding insertion.
Within the .INSERT-.ENDI pair, it is illegal to use
.EDIT, .MODULE, .REINSERT or .REMOVE pseudo-ops.
8. .ENDE - This pseudo-op delimits the range of an
edit. When this pseudo-op is seen, checks are made for
undefined labels, etc. and next edit or EOF can
be handled.
9. .REMOVE edit1,edit2,edit3...
This pseudo-operator removes the specified edits
from the selected module. Actually, the
only action taken is to put back the
original instructions displaced by the jumps
to the patch area for each .INSERT. (See appendix)
Specifically, no changes are made to the symbol table.
10. .REINSERT edit1,edit2,edit3...
For each specified edit, the effect of any previous
.REMOVE pseudo-op on that edit is reversed and the edit
is made active again.
Since the MAKLIB assembler is one pass, forward
references to labels and expressions are restricted to simple
addition and subtraction on the halfword boundary.
Essentially, references to as yet undefined labels or symbols
are legal where references to EXTERNAL symbols would be legal
in MACRO-10 (with no polish fixups). Literals are treated as
forward references, since the actual location of the literal
is not known until the .INSERT is ended. It is not legal to
define a label inside a literal. An additional restriction is
that the quantity used in the right-hand side of an assignment
(I.E. =,==,==: etc.) must not be forward or EXTERNAL.
Page 8
Assignments do not have to be inside .INSERTS in the fix file,
however the .EDIT and .MODULE pseudo-ops must precede any
assignments (which define new symbols in the symbol table).
Since it is impossible to backtrack references to a symbol in
the relocatable binary file, MAKLIB will not allow any
re-definitions of existing symbols. Therefore, any label or
symbol created by use of the FIX switch must be new to the
program.
For ease of patching, and to keep the appearance of the
binary patches as close to source level as possible, the
following pseudo-ops are implemented in the MAKLIB FIX file
assembler, an behave as they do in MACRO-10:
ASCII ASCIZ BYTE
COMMENT DEC EXP
IOWD OCT POINT
RADIX RADIX50 REMARK
SIXBIT SQUOZE XWD
NOTE
The pseudo-ops BYTE, DEC, OCT and EXP are limited to
generating one word of data at most.
All the MACRO operators and qualifiers are available as are the
defintions for the usual mnemonics such as CALLIs, UUOs, TTCALLs and
MTAPEs.
Symbols may be followed by the "##" (double pound sign) to
indicate that they are EXTERNAL quantities. However, if the symbol is
already in the symbol table, defined as EXTERNAL, the use of "##" is
not necessary. It is not necessary to follow undefined symbol names
with "#" (single pound sign), as any undefined symbol is assumed to be
a forward reference. If a symbol name is followed by the "#", the
only action taken is to give an error message if the symbol already is
defined. Labels may be defined as GLOBAL (available to other
programs) if they are followed by "::" (double colon), and more
generally, the full facilities available in MACRO-10 for combinations
of DDT supression and GLOBAL declaration are available for both labels
and assignments.
EXAMPLE:
OUT.REL=MASTER.REL,MASTER.FIX/FIX
2. WHO:iii - This switch is used only in conjunction with the FIX
command switch. It may appear as part of either the MASTER or the
TRANSACTION file specification. It takes as its argument up to three
alpha-numeric characters which are the initials of the person using
MAKLIB. These initials, if present, are included in the TRACE block
of any edits, in the field of who last affected any edit and in the
Page 9
field of who installed the edit. This switch could most conveniently
be placed in the user's SWITCH.INI file, so that placement of the
users's initials into the TRACE blocks would be an automatic process.
This switch is ignored if the command switch used is not FIX.
EXAMPLE:
OUT.REL=MASTER.REL,MASTER.FIX/FIX/WHO:ILG
Page 10
APPENDIX 1
Examples of FIX Files
.EDIT 345 ;THIS IS EDIT 345
.NAME ILG ;PATCH BY ILG
.DATE 3-DEC-75 ;PATCH WRITTEN DATE
.MODULE MACRO ;NAME FROM TITLE
.ASSOCIATED 200,-333 ;EDIT 200 REQUIRED,
;EDIT 333 PRECLUDED
REMARK THIS EDIT FIXES Q ERRORS WITH COMPLEMENT OPERATOR
.REMOVE 312 ;EDIT 312 SUPERSEDED
.INSERT EVUPAR+3,BEFORE,<PUSHJ P,MREAD> ;INSERT @EVUPAR+3
;BEFORE THE "PUSHJ"
;INSTRUCTION AT EVUPAR+3
;IS ""PUSHJ P,MREAD"
MOVEI AC0,CSTATN ;PICK UP CHAR STATUS
TRNE AC0,CH.FOO ;IS IT SPECIAL?
JRST [SETOM ERSTAT ;YES,FLAG IT
MOVEI AC0,CSTATX ;CHANGE STATUS
JRST .+1 ] ;RESUME NORMAL FLOW
.ENDI ;END OF FIRST PART
.INSERT XCTOP-1,REPLACE:3,<JRST XTC1A> ;REPLACE 3
;INSTRUCTIONS AT
;LABEL XCTOP-1
;WITH THIS CODE:
PUSHJ PP,MREAD1 ;USE 1ST CASE
SKIPE AC1,OPNUM ;ANY OP TO DO?
JRST @[ OP1
OP2
OP3]-1(AC1) ;YES,GO TO IT
JRST NOP ;NO,NO OP
.ENDI ;END SECOND PART
.ENDE ;END OF EDIT 345
Page 11
Another Sample Fix File
COMMENT $
THIS IS AN EXAMPLE OF A .FIX FILE USED TO APPLY TWO EDITS
TO A FILE. IT USES SOME FEATURES OF MAKLIB'S BINARY PATCHING
FACILITY.
$
REMARK THE FIRST EDIT DOES TWO INSERTS AND DEFINES A NEW SYMBOL
.EDIT 300 ;EDIT 300
.MODULE CBLIO ;SELECT 'CBLIO'
.ASSOCIATED 271 ;REQUIRES EDIT 271
.NAME ILG ;PATCH BY ILG
.DATE 30-DEC-75 ;ON 30TH DECEMBER
FL.JEM==FL.RAN!FL.OPN ;DEFINE NEW FLAG
.INSERT OPNUUO+3,BEFORE,<PUSHJ P, KILLIO> ;1.INSERTION @OPNUUO+3
;2.BEFORE INSTRUCTION THERE
;3.WHICH IS "PUSHJ P,KILLIO"
TRNE F,FL.JEM ;CHECK FOR THIS CASE
JRST NOFIL ;AND HANDLE IT
.ENDI ;END OF THE INSERT
.INSERT KILLIO+^D30,AFTER,<POPJ P,> ;INSERT NEW ROUTINE AFTER KILLIO
;1.@KILLIO+30.
;2.AFTER INSTRUCTION THERE
;3.WHICH IS A "POPJ P,"
NOFIL: ;NEW ROUTINE IS NOFIL.
PUSHJ P,.TCRLF## ;START NEW LINE
MOVEI T1,[ASCIZ "?NO FILE FOUND"] ;SET FOR MSG.
PUSHJ P,.TSTRG## ;OUTPUT STRING
JUMPN C,[MOVE T1,0(C) ;IF HAVE PTR TO FILENAME
PUSHJ P,.TSIXN## ;TYPE IT OUT
JRST .+1] ;AND RETURN
PUSHJ P,.TCRLF## ;NEXT LINE
POPJ P, ;END OF ROUTINE
.ENDI ;END OF INSERT
.ENDE ;END OF EDIT
REMARK THIS NEXT EDIT DELETES CODE AND REPLACES CODE
.EDIT 301 ;EDIT 301
.MODULE CBLIO ;SELECT MODULE FOR EDIT
.NAME ILG ;PATCH BY ILG
.DATE 31-DEC-75 ;DATE OF PATCH
.ASSOCIATED -265,300 ;EDIT 265 CAN'T BE PRESENT
;AND EDIT 300 MUST BE
Page 12
REMARK FIRST REPLACE SOME CODE
.INSERT RANIO+1,REPLACE,<TRNE F,FL.OPN> ;1.REPLACE STARTS W/INST @
;RANIO+1
;2.NEW INST. REPLACES OLD
;3.INSTR. TO MATCH IS
;"TRNE F,FL.OPN"
TRNE F,FL.JEM ;REPLACING INSTRUCTION
.ENDI ;END OF INSERT
REMARK NOW REPLACE 2 INSTRUCTIONS WITH 4 NEW ONES
.INSERT RAND-3,REPLACE:2,<MOVEI T1,"A"> ;1.INSERT @ RAND-3
;2.REPLACE 2 INSTRUCTIONS
;3.FIRST IS "MOVEI T1,"A" "
MOVEI T1,0(V) ;PICK UP STATUS
JUMPE T1,.POPJ## ;IF 0 , RETURN
CAIN T1,"A" ;IS IT ASCII "A"?
MOVEI T1,-" "(T1) ;YES,CONVERT TO SIXBIT
.ENDI
REMARK THIS LAST PART OF EDIT 301 SHOWS HOW TO DELETE CODE
.MODULE UFLOAT ;THIS PART OF EDIT
;301 IS IN MODULE UFLOAT
.ASSOCIATE -265 ;MAKE SURE 265 NOT PRESENT
;HERE EITHER.
.INSERT FLT.1+3,REPLACE:^D10,<MOVEM C,CURFLT> ;1.INSERT @FLT.1+3 WORDS
;2.REPLACE 10. INSTRUCTIONS
;3.THE FIRST OF WHICH IS
;"MOVEM C,CURFLT".
.ENDI ;REPLACE THE 10. INSTRS WITH
;NOTHING. I.E. DELETE THEM.
.ENDE ;END OF EDIT 301
Page 13
APPENDIX 2
Format of Trace Data Block (Link item type 1060)
The link item type, "TRACE BLOCK DATA" is used to include in the rel
file information that can be used to both verify and change the patch
status of a program. The format of the trace block follows:
The first part of the TRACE block is the static area. This area
appears in each module that is affected by the particular edit. The
static areas give information common to all modules affected by an edit
and the variable area gives the changing data on the particular edit as
it goes from module to module.
!=====================================!
TB$HED ! LINK ITEM NUMBER ! LENGTH OF BLOCK !
!-------------------------------------!
TB$EDT ! SIXBIT EDIT NAME (UP TO 6 CHRS) !
!-------------------------------------!
TB$STA ! -1 IF ACTIVE !WHO LAST AFFECTED !
!-------------------------------------!
TB$MAK ! WHO CREATED ! DATE (15 BIT) !
!-------------------------------------!
TB$INS ! WHO INSTALLED ! DATE (15 BIT) !
!-------------------------------------!
TB$FUT ! RESERVED FOR FUTURE USE !
!-------------------------------------!
TB$LEN ! # OF ASS. EDITS ! # OF PCO GROUPS !
!=====================================!
The static area, which is repeated in each module, is followed
by a variable area. The variable area consists of two parts, the first
giving data on the associated edit status for this module and the next
giving the actual program change orders (pco's). The length of each of
these areas appears in the static area of the trace block.
For each associated edit, the following group appears:
!=====================================!
TB$AEN ! SIXBIT EDIT NAME OF A.E. !
!-------------------------------------!
TB$AES !X! RESERVED FOR FUTURE ! 0B0 IF CAN'T BE PRESENT
!=====================================! 1B0 IF MUST BE PRESENT
After the associated edit groups appear, if there are any, the
pco groups for that module appear. There are currently three types of
program change groups: INSERT,REMOVE AND REINSERT. They can appear in
any order and the total number is of course variable.
INSERT program change order:
!=====================================!
Page 14
TB$PCO !PCO TYPE CODE (1) ! LENGTH OF GROUP !
!-------------------------------------!
TB$DAT ! 0 !INSTS INSRTD! ADDR. OF INSERT ! BITS 10-17
!-------------------------------------! ARE # OF INSTS
TB$PAT ! NEW ADDR OF ORG ! ADDR OF PAT CODE!
!=====================================!
REMOVE EDIT program change order:
!=====================================!
TB$PCO !PCO TYPE CODE (2) ! LENGTH OF GROUP !
!-------------------------------------!
TB$REN ! SIXBIT EDIT NAME !
!=====================================!
REINSERT EDIT program change order:
!=====================================!
TB$PCO !PCO TYPE CODE (3) ! LENGTH OF GROUP !
!-------------------------------------!
TB$RIN ! SIXBIT EDIT NAME !
!=====================================!
Page 15
APPENDIX 3
Format of Code Insertion
Insertion BEFORE a Location
.INSERT LOCATION, BEFORE, <ORIGINAL INSTRUCTION>
Generates:
LOCATION: JUMPA %PATCH
!-------------------------------------!
%PATCH: ! FIRST PATCH INSTRUCTION !
!-------------------------------------!
! SECOND PATCH INSTRUCTION !
!-------------------------------------!
/ /
/ /
/ /
!-------------------------------------!
! LAST PATCH INSTRUCTION !
!-------------------------------------!
! ORIGINAL INSTRUCTION !
!-------------------------------------!
! JUMPA 1, LOCATION+1 !
!-------------------------------------!
! JUMPA 2, LOCATION+2 !
!-------------------------------------!
! ANY "LITERALS" !
!-------------------------------------!
NOTE
The actual label created at the location
of the patched in code is of the form:
"%"<EDIT-NAME><EDIT-PART>
where the EDIT-PART is from "A" to "Z",
incremented for each .INSERT in the edit.
Page 16
Insertion AFTER a Location
.INSERT LOCATION, AFTER, <ORIGINAL INSTRUCTION>
Generates:
LOCATION: JUMPA %PATCH
!-------------------------------------!
%PATCH: ! ORIGINAL INSTRUCTION !
!-------------------------------------!
! FIRST PATCH INSTRUCTION !
!-------------------------------------!
! SECOND PATCH INSTRUCTION !
!-------------------------------------!
/ /
/ /
/ /
!-------------------------------------!
! LAST PATCH INSTRUCTION !
!-------------------------------------!
! JUMPA 1, LOCATION+1 !
!-------------------------------------!
! JUMPA 2, LOCATION+2 !
!-------------------------------------!
! ANY "LITERALS" !
!-------------------------------------!
Page 17
Insertion with REPLACEment at a Location
.INSERT LOCATION, REPLACE, <ORIGINAL INSTRUCTION>
Generates:
LOCATION: JUMPA %PATCH
!-------------------------------------!
! ORIGINAL INSTRUCTION !
!-------------------------------------!
%PATCH: ! FIRST PATCH INSTRUCTION !
!-------------------------------------!
! SECOND PATCH INSTRUCTION !
!-------------------------------------!
/ /
/ /
/ /
!-------------------------------------!
! nTH (LAST) PATCH INSTRUCTION !
!-------------------------------------!
! JUMPA 1, LOCATION+n !
!-------------------------------------!
! JUMPA 2, LOCATION+n+1 !
!-------------------------------------!
! ANY "LITERALS" !
!-------------------------------------!
NOTE
If no instructions are inserted (n=0),
then return is to LOCATION+1 and
LOCATION+2.
Page 18
Insertion with REPLACEment Instructions at a Location
.INSERT LOCATION, REPLACE:m, <ORIGINAL INSTRUCTION>
Generates:
LOCATION: JUMPA %PATCH
!-------------------------------------!
! ORIGINAL INSTRUCTION !
!-------------------------------------!
%PATCH: ! FIRST PATCH INSTRUCTION !
!-------------------------------------!
! SECOND PATCH INSTRUCTION !
!-------------------------------------!
/ /
/ /
/ /
!-------------------------------------!
! LAST INSTRUCTION OF PATCH !
!-------------------------------------!
! JUMPA 1, LOCATION+m !
!-------------------------------------!
! JUMPA 2, LOCATION+m+1 !
!-------------------------------------!
! ANY "LITERALS" !
!-------------------------------------!
NOTE
If m is not specified or is zero, the
effect is the same as the REPLACE keyword
without an argument, I.E. one word is
skipped over on return for every one
inserted.