macro language and the macro processor The assembly language programmer often finds it Necessary to repeat some blocks of code many times in the course of a program. The block may consist of code to save or exchange sets of registers, for example, of code to set up linkages or perform a series of arithmetic operations. In this situation the programmer Gia instructions (often called will find a macro instruction facility-useful. ine a ions for grou macro, the programmer essentially defines a single “instruction” to represent a block of code. For every occurrence of this one-line macro instruction in his pro- gram, the macro processing assembler will substitute the entire block.» By defining the appropriate macro instructions, an assembly language pro- grammer can tailor his own higher level facility in a convenient manner, at no cost in control over the structure of his program. He can achieve the conciseness and ease in coding of high level languages without losing the basic advantage of assembly language programming. Integral macro operations simplify debugging and program modification and they facilitate standardization. Many computer manufacturers use macro instructions to automate the writing of “tailored” operating systems in a process called systems generation. Macro instructions are usually considered an extension of the basic assembler language, and the macro processor is viewed as an extension of the basic assem- bler algorithm. As a form_of programming language, however, macro instruction languages-differ significantly from assembly Janguages and compiled algebraic languages, Important-analogs are to be found in some high level languages and text editing systems. In this section we will discuss an assembler macro facility. We will treat the use of macro instructions in programming and the implementation of macros within an assembler. Certain features that are characteristic of most macro processors will be introduced and developed. wtMACRO Nap, U 12 x 1 iS 4.1. MACRO insTRUCTION! a macro is an In its simplest form, abbreviation for a seq . s simplest forn ior the Following PORT n' Of 6, Consi ce Example 1: 1,DATA Add contents of DATA to register 4 A ‘Add contents of DATA to register 2 2, DATA aT eres , 4 i ‘Add contents of DATA to register 3 , A 3, DATA 4 ‘Add contents of DATA to register 1 ‘ Hy a ‘Add contents of DATA to register 2 a 3,DATA Add contents of DATA to register 3 A , DATA O0C FS’ Inthe above program the sequence 1,DATA 2,DATA 3, DATA \ ee Occurs twice. A macro facility permits us to attach a name to this sequence and to use this name in its place. We can invent a macro language that allows usto specify the above as a macro definition and allows us to refer to ‘the definition later. We have chosen to use a 360-type macro language. _ A.macto processor effectively constitutes a separate language processor with itsown language, / Me attach a name to a sequence by means of a macro instruction definition, ‘Which is formed in the following mariner: ——— Start of definition ‘Sequence to be abbreviated a MEND fi ( 1 End of definition iacne LANGUAGE AND THE MACRO PROCESSOR 113 3 MACRO pseudo-op i: i Cie se ae op is the first line of the definition and identifies the fol- owi nacro instruction name. Following the name line is the of_instructi (i fi seu eon The being abbreviated—the instructions comprising the mai The definition is terminated by a line with the MEND (‘macro end’) pseudo-op. ong the macro has peso deh the_use of the macro name as.an. operation -an_assembly program is equivalent to the use of the corresponding mnemonic fnstruction sequ Our example might be rewritten as follows, assigning the ~~» ‘{NCR’ to the repeated sequence. name Source | Expanded source MACRO INCR A 1,DATA | & A 2:DATA 2 A 3,DATA | MEND INCR | A. 1,DATA ' A 2,DATA l A 3,DATA INCR A 4,DATA : A 2IDATA ; | ne 3,DATA DATA pe F's" | DATA pe” rs In this case the macro processor replaces each macro call with the lines A 4, DATA A 2, DATA [ke A 3.DATA the macro. Notice that the comentis called expanding. iat t ne defini- macro definition itself does not apPcay in the expan' tion is saved by the macro processor. The occurrence in th | the. macro name, as an_operation. mnemonic to be expande! a) : ded source code. T' ¢ source program of d, is called a macro This proce:414 FEATURES OF AMACRo », bg ‘ 4.2: FEATURES OF A MACRO FACILITY 4.2.1 Macro Instruction Arguments The macro facility presented thus far is capable of inserting blocks of tions in place of macro calls. Alll of the calls to any given macro wil] be rey by identical blocks. This macro facility lacks flexibility: there is no y,. specific macro call to modify the coding that replaces it. An important Xt, Of this facility consists of providing for arguments, or parameters, in mac, Corresponding macro dummy arguments will appear in macro defini sider the following program. Example 2: \ A 1,DATA1 A 2,DATA1 A 3,DATA1 A 1,DATA2 A 2,DATA2 A 3,DATA2 DATA1 DC Fo DATA2 DC F'10" In this case the instruction sequences are very similar but no identical! first sequence performs an operation using DATAI as operand: the second, DATA2. They can be considered to perform the same operation with avait Parameter, Or argument. Such a parameter is called a macro insiruction’. ment, or “dummy argument;” it is specified on the macro name line aid! tinguished (as a macro language symbol rather than an assembly languag bol) by the ampersand (&), which is always its first character. The pret! program could be written as:po LANGUAGE AND THE MACRO PRocesson mac! Source si Expand. MACRO Macro INCR has | ded source INCR BARG | One argument , A 1,8ARG A 2,BARG/ \_~ A 3,8ARG / MEND | INCR DATA Use DATAY | ; a Poesia {2 ian on oni | A 3DATAI INCR DATA2 Use DATA? operand | { a Lpataa : | A 3,DATA2 < FS’ DATAL DG a DATA1 pc FS? DATAz OC 10 jeara: BE ft : It is possible to supply more than one argument in a macro call. Each axpu- ment must cortespond to a definition (“dummy”) argument on the iiaero name line of the macro_definitionWhen a macro call is processed, the arguments sup- plied are substituted for the respective dummy arguments in the macro definition. = Consider the following program: Example 3: Loop! A 1,DATA1 A 2, DATA2 A 3, DATAS Loor2 A 1, DATA3 A 2, DATA2 A 3,DATA1 DATA1 DC FS! DATA2 DC F'10" DATA3 DC FIs"ao PERT UR EE OP RACKS Fa, V6 gre different, on are cence ee mn In thi ands in the compe : could be written as on a aa Bats of the sequences. THIS program EXPANDED a0, SOURCE RO \ twen BARG1 BARGZBARCS era 1A 1 ARG1 eA 2.&ARGZ A 3RKARG3 MEND | , sazpares | LooPt A Loop? «NCR DATA1,DA A \ A ‘ /DATAA \ LOOP2 A Loor2 INCR DATAS,DATA2 A \ a : Bi | DATA DC FS) ee | cares DATA3 DC F158" BATA eC : | flere we have specified four arguments, including arguments are treated just like those written as “operan tion; the argument appearing in the label field on the macro name calls to the macro differs from other arguments only in its location. used.as labels within the macro need not appear in the label field o arguments appearing in tl fhe label field of a call may be used asope the macro instruction code. Lab el arguments are convenient buts features that enhance the similarity between macro instructi ions ai assembly language instructions. Following is another encoding ofpo LANGUAGE AND THE MACRO PRocesson mac! "7 5 vac , Expanded source 3 INCR BARG1&ARG2, rt aoe SARGBLAR girs) 2,8&ARG2 — 3,8ARG3 | MEND INCR DATAI,DATA2,DATA3, , }LOOP1! {Loopy A 4.DATAL A 2oaTa2 : 3,DATAS incr DATAS.DATA2,DATA1,LOOP2, (Loop? 1DATAZ | A 2.DATA2 A 3,DATAT There are generally two_ways-of specifying arguments to a macro call The fust, positional argiment, is demonstrated inthe preceding examples: Arsw. ments are matched with dummy arguments according to the order in which they appear. Thus, in a macro call INCR A,B,C’ A, B, and C replace the first, second, and third dummy arguments. A more general argument facility, keyword ae ments, allows reference to dummy arguments by name as well as by position. Referring to the dummy arguments appearing in the definition of INCR, we might write &ARG2=B! or ‘INCR &ARG 1=A,&ARG2=BARG3=C’. ‘INCR &ARG1=A, &ARG3=" This can be useful when a macro has arguments fabels, for example—that are not always needed. Any arguments not supplied are presumed_blank by the macro processor. 4.2.2, Conditional Macro Expansion Two important macro processor pseudo: reordering of the sequence of macro-expansion. of the machine instructions that appear in expansi Consider the following program: yp, AIF and AGO, permit conditional This allows conditional selection ions of a macro call.Example 4: Loop: a a 1,DATAL a 2, DATAZ 3, DATAS Loor2 a 1, DATAZ a 2, DATA2 Loops A 1,DATAI DATA1 DC 5 DATA2 0c F’10" DATA3 DC F’15" ands, labels, and the number of instructions gener In this example, the oper: Id be written as follows: change in each sequence. This program cou. MACRO &ARGO VARY Be &ARG1,&ARG2,8ARG3 &ARGO A 1,8A) AIF (écounr EQ 1).FINI Test if & COUNT = A 2,8ARG' AIF FE SUNT EQ 2).FINI Test if & COUNT = A 3,8ARG3 oa oS FINI MEND Expanded source Loop! VARY BDATALDATA2,DATAS | { LOOPt =A LDAT: A | A 30aT) Loop2 - VARY_~—s2,DATA3,DATA2 Loop2 ADT : | A 20a0 Loops VARY 1,DATA1 | | Loops = ASCANAT DATA1 DG F's! DATA2 OC F'10" DATA3 OC F15'Quack AND THE MACRO PROCESS Oo . pret Pi with a period (.), such as FIN}, 574 me Meg js HT giput of the Macro processor. The Res ety Pees a) Ute ance 4 eee eet colt are fie) date ynesponding (© RCOUNT isa tome ua ame ¢ with the statement following the Alp Phew + the macro process, jae With | an do-op, e ‘conti conditional branch pseudo-op, it performs an . th PAF iS ir the tested condition is true, The agg she fst ant ies oop or “go 10" statement, 10 specifies a label sn ational wement in the macro instruction definition; the a a ae ner ential processing of instructions with the indicated state jos a sre directives. 10 the macro processor and do not ‘apes Pee Just 3s conditional and unconditional branch icatro ms git 3 mstructions in a SONS — age program direct the order of program flow, AIF and Aco n * equenc in which the macro processor expands the statements in ma cag. This makes it possible to tailor specific occurrences of in iets. Fe i = er to their various contexts. If machine language branches and tests are en y will waste execution time; also, the tests and unused instructions wil in the macro instruction language per- at assemble selectively, and by testing TATE (CG et labetted Fy we CESS con. fement é of ae 5 i. call, omit code not needed. This facility for selective assem- hi shly general macros is one of the systems programmer's most powerful saming tools. 423 Macro Calls Within Macros : cro calls are “abbreviations” of instruction sequences, it seems reason- z that such “abbreviations” should be available within other macro defini- tions. For example, Sample 5: MACRO ADDI &ARG L 1, &ARG A 1, =F ST 1, &ARG MEND MACRO ADDS &ARG1, BARG2, &ARGS A0D1 &ARGI me &ARG2 uenp BARG3 i sparate calls 10 ° Oe tin the definition of the macro ‘ADDS’ are three 2