Mac-Source 1994 July

home *** CD-ROM | disk | FTP | other *** search

/ Mac-Source 1994 July / Mac-Source_July_1994.iso / Modula / Information / GUIDE.TEXT next >

Wrap

Text File | 1985-05-03 | 44KB | 1,417 lines

User Guide to the Modula-2 System --------------------------------- Table of Contents 4. MODULA-2 Compiler, Linker, Debugger 4.1. A User Guide to the MODULA-2 System 4.1.1. Glossary and examples 4.1.2. (** The M3 MODULA-2 base **) 4.1.3. Compiling 4.1.4. Linking 4.1.5. Running a program 4.1.5A Call by command interpreter 4.1.5B overlay organization 4.1.5C call from a program 4.1.5D run time support for program management 4.1.5E (** converting from .LOD to .IMA format **) 4.1.6. Debugging 4.1.7. Library Searching Strategies 4.1.8. The module SYSTEM 4.1.8A Workfiles 4.1.8B Commandfiles 4.1.8C History (Acknowledgements) 4.1.9. Compiler error messages (MODLIB:GUIDE.TEXT 23.04.85) 4. MODULA-2 Compiler, Linker, Debugger -------------------------------------- adapted by Institut fuer Elektronik. Version 5.8.1983, prepared by P. Fink ETH Zuerich. In diesem Kapitel erklaeren wir detailliert die MODULA-2 Programmierumgebung . Da viele der Prinzipien des PDP-11 MODULA gelten, wurde die Originaldokumentation "A User Guide to the MODULA-2 System" als Grundlage gewaehlt (in englischer Sprache). (* parts which are not valid for the portable MC68000 Compiler, have been put into comment brackets by P. Fink.(**...**) Some remarks about the SMAKY8 have been added when necessary *) 4.1 A User Guide to the MODULA-2 System --------------------------------------- The Modula system consists of a compiler, a linker, a debugger and possibly some support utilities. the compiler and linker are themself modula programs. On the SMAKY 8, a modula program is simply called by typing X progname in the CLE line. Therefore, the compiler is called by X modula . The linker is called by X linker . A program may be stopped with the CLEAR key on the numerical pad. This works almost anytime, except in tight loops. The CLEAR key is checked on every system request. (** MODULA-2/68K is a resident programming environment for MODULA-2 that runs on the MC68000 based "M3" system. The software consists of a compiler, a linker, a debugger and utility modules, which are all linked to the operating system. The operating system first gives control to the command interpreter, which accepts the file name of the next program to be executed. The command interpreter then terminates and calls the loader, which is also embedded in the operating system. It loads the named file and transfers control to that program. Upon termination of the program, control returns to the operating system, and the command interpreter asks for the next program. In an error situation control returns to the operating system, too. The exception handling part has to correct the system state (e.g. correct the stack pointer). Therefore at the command interpreter level the system is always in a consistent state. **) Compiler, linker, and debugger are Modula-2 programs. They are treated by the operating system like any other user program. The following chapters require a sound knowledge of the language itself. Reference literature: N.Wirth: Programming with MODULA-2, Springer Verlag. 4.1.1. Glossary and examples ---------------------------- compilation unit : Unit accepted by compiler for compilation, i.e. definition module or program module (see Modula-2 syntax). definition module : Part of a separate module specifying the exported objects. program module : Implementation part of a separate module or main module. source file : Input file of the compiler (compilation unit) (default extension MOD). symbol file : Compiler output file with symbol table information; information generated during compilation of a definition module (default extension SYM). reference file : Compiler output file with debugger information, generated during compilation of a program module (default extension REF). link file : Compiler output file with code in Modula-2 linker format (default extension LNK). load file : Linker output file with code in Modula-2 loader format (default extension LOD). map file : Linker output file with storage layout information (default extension MAP). (* system dependent definitions: specific for the SMAKY 8 *) filename : File names are at most 11 characters long (plus device, plus extension). When a module name is longer, then its corresponding filename is truncated to 11 characters. default device : a device name needs not to be specified on the SMAKY 8. default extension : when a filename doesnt contain an extension or is terminated with a point, then this default extension is appended to the name default libraries : on the SMAKY8, the default libraries searched are the directory of the source file, then <empty>, i.e. #MM0:, and then MODLIB: Examples: --------- The following examples are used in the description: MODULE Prog1; ... END Prog1. MODULE Prog2; BEGIN a := 2 END PROG2. DEFINITION MODULE Prog3; EXPORT QUALIFIED ... ... END Prog3. IMPLEMENTATION MODULE Prog3; ... END Prog3. DEFINITION MODULE Prog4; IMPORT SystemTypes; EXPORT QUALIFIED ... ... END Prog4. IMPLEMENTATION MODULE Prog4; IMPORT SystemTypes, Loader, Exceptions; ... END Prog4. MODULE Prog5; IMPORT Prog3, Prog4; ... END Prog5. (** 4.1.2 The M3 MODULA-2 base -------------------------- The operating system is loaded and started when the M3 system is booted (e.g. after GO23000). Immediately the command interpreter is executed. The command interpreter displays the prompt symbol "->" to indicate that it waits for a command (which can result in a utility program call). Normally a user program can be interrupted by twice <ctrl>C; then control returns to the operating system. **) 4.1.3 Compiling --------------- Compilation of a program module: -------------------------------- The compiler is called by typing X MODULA, in the CLE line. After displaying the string "source file>" the compiler is ready to accept the filename of the 'compilation unit' to be compiled. Default device : -- Default extension : MOD ->X MODULA<cr> source file> PROG1<cr> (* name PROG1.MOD is assumed*) p1 p2 (* indicates succession of *) p3 (* activated compiler passes *) p4 p5 end compilation -> If syntactical errors occur in the compiled unit, compilation stops after the third pass and a listing with error messages is generated: ->X MODULA<cr> source file> PROG2<cr> p1 ---- error (* error detected by pass 1 *) p2 p3 ---- error lister end compilation -> Compilation of a definition module: ----------------------------------- For definition modules the filename extension DEF is recommended. The definition part of a module must be compiled prior to its implementation part. A symbol file is generated for definition modules. ->MODULA<cr> source file> PROG3.DEF<cr> (* definition module *) p1 p2 symfile lister end compilation -> Symbol files needed for compilation ----------------------------------- At the compilation of a definition module, a symbol file containing symbol table information for the compiler is generated. This information is needed by the compiler in two cases: - At compilation of the implementation part of the module. - At compilation of another unit, importing objects from this separate module. According to a program option set at the beginning of the compilation (Q option), the compiler asks for the needed symbol files or attaches them by a default name (the first n characters of the module name, see glossary). ->X MODULA<cr> source file> PROG3<cr> (* implementation module *) p1 Prog3: DK:PROG3.SYM p2 p3 p4 p5 end compilation -> ->X MODULA<cr> source file> PROG4.DEF/QUERY<cr> p1 SystemTypes> SY:SYTY<cr> p2 symfile lister end compilation -> ->X MODULA<cr> source file> PROG5/QUERY<cr> p1 Prog3> PROG3<cr> Prog4> PROG4<cr> p2 p3 p4 p5 end compilation X MODULA source file> PROG6/L library prefixes> D:,XY: p1 ..... -> Files generated by the compiler: -------------------------------- Several files are generated by the compiler. They all obtain the same device:filename as the source file with the following extensions: for definition modules : listing *.LST symbol file *.SYM for program modules : listing *.LST reference file *.REF code file *.LNK Program options for the compiler -------------------------------- When reading the source file name, the compiler also accepts some program options from keyboard. Program options are marked with a leading character '/' and must be typed sequentially after the file name. The compiler accepts the options: /QUERY : Compiler explicitly asks for the names of the needed symbol files belonging to modules imported by the compiled unit. /Q is a short form for this option. /NOQUERY : No query for symbol file names. The files are searched corresponding to a default strategy. /LISTING : A listing file must be generated. /LIST is accepted for this option. /NOLISTING: No listing file must be generated. /N is a short form for this option. /EL : A listing will only be generated when there are compilation errors. Note all errors are also displayed on the terminal. /VERSION : Compiler has to display information about its version. /V is a short form for this option. /LIBQUERY : library query: the compiler has to ask for the libraries it should search when looking for symbol files. Up to 5 user libraries may be specified, separated by comma. Note: <empty> and MODLIB: are always searched in addition to these 5 libraries and need not to be specified. Normally, the directory of the source file, <empty> and MODLIB: are searched. /L is an abbreviation for /LIBQUERY. /LARGE : Large program option. Normally the compiler allocates tables and files for a small to medium sized program. This option forces it to allocate large tables and files. /LA is an abbreviation for /LARGE. Defaults: * NOQUERY and EL are set as default options. Compilation options in compilation units ---------------------------------------- Comments in a Modula-2 compilation unit may be used to specify certain compilation options for tests. The following syntax is accepted for compilation options: Options = Option { "," Option } . Option = "$" Letter Switch . Switch = "+" | "-" | "=" . Options must be the first information in a comment clause. They are not recognized by the compiler, if other information preceedes the options in the clause. Options : T Array index and case label boundary test. S Stack overflow test. P No generation of procedure entry and exit code. C CLR option. C- means the compiler must not generate CLR instructions. Switches: + Test code is generated. - No test code is generated. = Previous switch becomes valid again. Defaults : All switches are set to "+" by default. Example : MODULE x; (* $T+ *) ... test code generated ... (*$T- *) a[i] := a[i+1]; no test code is generated (*$T= *) ... test code is generated ... END x Remark concerning the C option: This option is useful when writing device drivers. Since the CLR instruction of the MC68000 does a READ/WRITE cycle, it should not be generated for absolute variables which are device registers (e.g. UADATA [300h]: CHAR;) When such a variable is passed to a procedure as an argument, then within the procedure, the compiler doesn't know it is an absolute variable and may erronously generate the CLR instruction. With C- in the procedure, it is instructed to generate a MOVE #0,... instead. Module key: ---------- To each compilation unit the compiler generates a so called "module key". This key is unique and is needed to distinguish different compiled versions of the same module. The module key is written on the symbol file, the link file and the load file. For an implementation module no new key is generated. It gets the same key as the definition module. The module keys of imported modules are also recorded on the symbol files and the link files. Any mismatch of module keys belonging to the same module will cause an error message at compilation or link time. CAUTION : Recompilation of a definition module will produce a new symbol file with a new module key. In this case the implementation module and all units importing this module directly or indirectly must be recompiled. Recompilation of an implementation module does not affect the module key. Differences and restrictions in implementation: ---------------------------------------------- For the implementation of Modula-2 for MC68000-based computers some differences and restrictions must be considered. FOR statement ------------- The control variable must not be of type CHAR or any enumeration type (byte size), if step is not -1, or 1. The step must not be greater than 77777B. CASE statement -------------- The labels of a case statement must not be greater than 77777B. Dynamic arrays -------------- In a procedure declaration, only the form VAR ARRAY OF <anytype> is available, but not ARRAY OF <anytype> . For this reason, the compatibility rules between formal and actual parameters have been lowered for VAR ARRAY OF CHAR and VAR ARRAY OF WORD: - strings are considered compatible with VAR ARRAY OF CHAR. - everything is compatible with VAR ARRAY OF WORD. Type Transfer Functions between types of unequal size ----------------------------------------------------- The type transfer functions to a shorter size (e.g. CHAR(cardinal)) transfer the lower order bits of the source. Type transfers to a longer size transfer the 'value'. For unsigned source types, zeroes are filled into the higher bits (e.g. ADDRESS(char), ADDRESS(cardinal)). For a signed source type, the sign is extended (e.g. ADDRESS(integer)); PROCESS, NEWPROCESS, TRANSFER, IOTRANSFER, LISTEN, SYSRESET, module priorities. ------------------------------------------------------------ These language concepts for concurrent processing are implemented. SMAKY 8 notes: The implementation assumes the tasks are running in User Mode. All processes are part of the same NTREL task. PROCESS, NEWPROCESS and TRANSFER may be used freely, even for structuring application programs. Before using IOTRANSFER (which handles peripheral interrupts), it is recommended to consult the actual implementation listings. IOTRANSFER may interfere with the operating system PSos. Compatibility with SMILERX V4 ----------------------------- Symbol Files and Link Files produced by SMILERX V4 are acceptable when transferred properly from a CYBER to the SMAKY 8 and converted. MC68000 Modula-2 compiler profile --------------------------------- INTEGER : -32768..32767 CARDINAL : 0..65535 ADDRESS : 0..4294967295 REAL : -3.4028232E38..-1.1754943E-38, 0.0, +1.175494E-38..+3.4028232E38 (IEEE 32-bit Format) Max. Index : -32768..65535 Max. Range : 32767 ARRAY size : 32766 max. RECORD size : 32766 max. Global Variables: 32766 max. Allocation of Modula-2 types: BOOLEAN : Byte (1) CHAR : Byte (1) INTEGER : Word (2) CARDINAL : Word (2) REAL : Long (4) WORD : Word (2) ADDRESS : Long (4) BITSET : Word (2) pointer : Long (4) enum. ( <= 256 elem. ) : Byte (1) enum. ( > 256 elem. ) : Word (2) set ( <= 8 members ) : Byte (1) set ( > 8 members ) : Word (2) 4.1.4 Linking ------------- Compiler code output files (called link file, with extension LNK) must be linked before run time. One or more link files can be linked to a program. The generated code file (called load file, with extension LOD) may be loaded by the loader. Call the linker by typing X LINKER in the CLE line. After displaying the string "master file>" the linker is ready to accept the filename of the master file (main program). ->X LINKER<cr> master file> PROG1<cr> (* PROG1.LNK is assumed *) The linker generates two files with the same name as the master file and the following extensions: LOD : Load file with code to be loaded. MAP : Map with allocation addresses of linked modules (only when program option M is set). A program (overlay) may be linked to an already linked base. Information about this base is needed by the linker. To get this information, the linker has to attach the load file (extension LOD) of this base. (** Normally the base is the resident part of the basic executive. The file SY:MODULA.MXS is the default base file attached by the linker, if no other base file is specified (see program option B). **) All separate modules imported by the master module (or other linked modules) must be linked. If a needed module is not already linked (e.g. in the base), then the linker requests the corresponding link file. To search this file a default strategy or a query strategy can be chosen (see program option Q). There is also an option the change the libraries for searching (see option /L) When reading the name of the master file the linker also accepts some program options from keyboard. Program options are marked with a leading character '/' and must be typed sequentially after the file name. The linker accepts the options: /B : Linker has to ask for the name of the base file. /Q : Linker has to ask for the names of link files belonging to imported modules. If Q is not set, then the files are searched corresponding to a default strategy. /L : Linker has to ask for the names of the libraries to be searched. Up to 5 user libraries may be specified, separated by comma. In addition to these libraries, <empty> and MODLIB: will always be searched when looking for imported modules. Normally, the directory of the master file, <empty> and MODLIB: are searched. /M : A map file must be generated. /D : Drop Tables Option: The Linker is inhibited from writing certain information to the load file. The load file becomes shorter. The following two tables are omitted, and the consequences are: a) the Linker Tables. Overlays cannot be linked above this program. b) the Debugger Tables. The symbolic Debugger cannot be used for this program (unless linked again without the /D option). /V : Linker has to display information about its version. (**S : System will be generated, i.e. there is no base to link to it. A : Use only with option S. Linker has to ask for the start (lowest) address of the new generated system. If A is not set then a default value (40000H) is substituted to the start address. **) ->X LINKER<cr> master file> PROG4/Q<cr> (* base is the operating system *) end linkage (* imported modules are already *) -> (* linked in the resident part *) ->X LINKER<cr> master file> PROG5/B/Q/M<cr> base file> PROG4<cr> (* DK:PROG4.LOD is accepted *) files linked: Prog3> PROG3<cr> (* DK:PROG3.LNK is accepted *) end linkage (* DK:PROG5.MAP is generated *) -> 4.1.5 Running a program ----------------------- A linked program is ready to be called (i.e. loaded and executed). There are two different possibilities to call a program: - pass a file name to the command interpreter. - activate the procedure Call, exported from module SM8Loader. (by a program) (** 4.1.5A Call by command interpreter ---------------------------------- A program linked to the resident part of the basic executive (MODULA.MXS) is called by passing its filename to the command interpreter. After displaying an "->" the command interpreter is ready to accept a file name. For the search strategy refer to (2.2.2 Utility-Programme). If an error occurs during loading or execution, then the command interpreter displays an error message. If enabled, a dump of the core image will be written on file SY:DUMP.COR after an execution-error. (refer to 2.1.2 Command DON/DOFF) ->PROG1.LOD<cr> PROG1.LOD is loaded and executed ... ->PROG5.LOD<cr> ---- wrong load key program not linked to operating system -> **) 4.1.5B Overlay organisation: --------------------------- A so called overlay organisation is implemented in the Modula-2 system. It is a simple instrument to overlay program parts. An overlay part is a load file, produced by the linker. It consists of one or more linked modules. An overlay part is loaded and executed, when it is called from that base part, to which it has been linked (the base file). This simple rule allows to have program overlays on several levels. (**In the Modula-2 system the operating system is the first overlay level. All programs started via command interpreter are running on the second overlay level. **) Example : Storage layout by code during execution of a program storage ------------------------------> | | B is called from A | |------------- | | B | B is finished | |------------- | | C is called from A | |-------- D is called from C | | |------------ | | | | | A | | D | | | | | D is finished | | C |------------ | | | E is called from C | | |-------- | | | E | E is finished | | |-------- C is finished | |-------- | | 4.1.5C Call from a program: -------------------------- If a program P2 is not linked to the basic executive, then it must be called directly from the base part P1, to which it has been linked, i.e. the procedure Call, exported from module Loader, must be activated (see DEFINITION MODULE Loader). 4.1.5D Run time support for program management: ---------------------------------------------- Some functions for abnormal termination and for program recovery are available from the module Exceptions. (**If P2 terminates by an error, then it is recommended to propagate the error from P1 to the basic executive to cause a core image dump and reinstallation of the underlaying operating system (see DEFINITION MODULE Exceptions). Example : Fragments from module PROG4. VAR loadres : SystemTypes.LoadResultType; exerror : SystemTypes.ErrorType; ... Loader.Call("DK PROG5 LOD",loadres,exerror); IF exerror <> SystemTypes.NormalReturn THEN (* cancel execution *) Exceptions.PropagateError; ELSE ... **) (** 4.1.5E Converting from .LOD to .IMA format ------------------------------------------ The M3 utility program LODIMA allows to transform the .LOD output file produced by the linker into a fast loadable format .IMA (refer to 2.5). **) 4.1.6. Debugging ---------------- If a run time error occurs and a program terminates irregularly, then -if switched on- the Modula-2 system writes the current image of the memory onto the dump file DUMP.COR . This 'post mortem dump' can be inspected with the debugger. Starting the debugger --------------------- To run the the debugger type X DEBUG in the CLE line. The debugger first asks for two file names: the dump file and the load file of the crashed program. Yyou may hit <RETURN> twice and accept the defaults. For the dump file, the default is DUMP.COR, and for the load file, it is the load file name of the crashed program, taken from the dump file. With the load file name, you may set the query or libquery option to steer the access to the reference and source files. The options which may be given with the load file name are: /Q query option. DEBUG should ask for the names of the reference and source files (REF and MOD) /L library query option. DEBUG should ask for the names of libraries to be searched for REF and MOD files. When no options are given, then REF and MOD files are searched automatically of the library of the load file. Example: ->X DEBUG<cr> dump file> DUMP.COR crashed program XY:PROG1.LOD load file> XY:PROG1.LOD or load file> XY:PROG1.LOD/L libraries> A:,AB: ... generating P-chain Prog1.REF> XY:PROG1.REF<cr> Files of the debugger --------------------- The debugger needs the following files: - the dump file, DUMP.COR. - the load file of the crashed program - the .REF and the .MOD files of the modules to be debugged. The dump file must be created explicitly be the user, otherwise no dumps can be taken. Windows of the debugger ----------------------- The debugger shows the crashed program through five windows (one at a time): P(rocess window : shows the active procedures at the moment of the dump. Also, the procedure chains of other PROCESSes may be shown. M(odule window : shows the list of all linked modules. After choosing a module, the T and D window will display data of the selected module rather than the procedure selected in the P window. D(ata window : shows the value of the variables of the active procedures and the value of the global variables of the modules. Even local modules, structured variables and pointer chains may be inspected. T(ext window : shows the source file, located at the point in error. For the T window, the debugger needs the source file and the reference file, which are searched or asked for (depending on options). Example: PROG1.MOD> XY:PROG1.MOD <cr> C(ore window : shows an absolute memory dump. Commands of the debugger (DEBUG V1D) ------------------------------------ window switching commands: "P": switch to P window "M": switch to M window "D": switch to D window "T": switch to T window "C": switch to C window "Q": quit from debugger window specific commands: P window: "@": re-generate the P-chain of the process which was running when the program was aborted M window: none D window: "S": go to the son level (down) "F": go back to the father level (up) "V": go back to the last level with variables "A": show address of the currently selected data element "@": generate the P-chain of the currently selected variable or field of type PROCESS or POINTER TO PROCESS, or of a PROCESS variable specified by its address and switch to the P window T window: none C window: "A": ask for a new address and show its environment in the core "@": take the currently selected word as a new address (indirect addressing) changing the data representation type: "#C": CHAR "#B": byte (octal) "#W": WORD (octal, default) "#I": INTEGER "#U": CARDINAL (unsigned) "#R": REAL general commands (legal in all windows): change position: "+",<ESC>"B" (VT52/VT100 down arrow): increment position "-",<ESC>"A" (VT52/VT100 up arrow): decrement position "^": increment position by half a screen length "_": decrement position by half a screen length number: set a new window position (line number) terminal: <ESC>"H": hardcopy terminal <ESC>"V": VT52/VT100 terminal (default) 4.1.7 Library Searching Strategies ---------------------------------- The following strategies are applied by the compiler and linker to search the needed symbol or link files. The default strategy or the query strategy can be chosen. Both strategies refer by default to the default libraries (source or master file directory, <empty> and MODLIB:). A default name is generated from the module name. If the module name is longer than 11 characters, then the first 11 characters are taken for the file name (SMAKY 8). Default strategy ---------------- All needed files are searched by the default name (11 leading characters of the module name). A file is searched in the same directory as the source/master file first. If this search fails, then the file issearched on devices #MM0: and MODLIB:. For each file a message is displayed, where is was found or that it was not found. In the latter case the user has no possibility to type another file name. The libraries to be searched by the default strategy may be changed for the compiler and the linker by the /L options. Query strategy -------------- For each searched file the user has to type a file name (default device is <empty>). If search fails, then an error message is displayed and another file name must be typed. If only <cr> is typed, then the file is searched as in the default strategy. Query for a file name is repeated until the file is found or <esc> is typed. Typing <esc> means, that no file is supplied. On the SMAKY8, <esc> means the END key, <can> means the UNDO/ESC key. 4.1.8 The module SYSTEM ----------------------- The module SYSTEM offers further tools of Modula-2. Most of them are implementation dependent and/or refer to the given processor. Such kind of tools are sometimes necessary for the so called "low level programming". SYSTEM contains also types and procedures which allow a very basic coroutine handling. The module SYSTEM is directly known to the compiler, because its exported objects obey special rules, that must be checked by the compiler. If a compilation unit imports objects from module SYSTEM, then no symbol file must be supplied for this module. Objects exported from module SYSTEM for this 68K Version: Types ----- WORD Representation of an individually accessible storage unit (one word = 16 Bit on MC68000). No operations are allowed for variables of type WORD, except assignment. A WORD parameter may be substituted by an actual parameter of any type that uses one word in storage. If the parameter is a value parameter, then values of types using one byte are allowed too for substitution. The same holds for the substitution of a dynamic ARRAY OF WORD parameter by a value of a type with an odd size. ADDRESS Byte address of any location in the storage. The type ADDRESS is compatible with all pointer types and is itself defined as POINTER TO WORD. All cardinal arithmetic operators apply to this type, as well as INC and DEC. ADDRESS may be used in the sense of "long cardinal (32bit)". The type ADDRESS is not compatible with the type CARDINAL. Type transfer functions (zero-filling resp. truncating) have to be used in mixed arithmetic expressions. Under the $T- option, INC and DEC are allowed for any pointer variables, not only for ADDRESS. PROCESS Type used for process handling. (points to the (moving) top of the process stack) Procedures ---------- NEWPROCESS(p:PROC; a: ADDRESS; n: ADDRESS; VAR p1: PROCESS; ip: ADDRESS) Procedure to instantiate a new process. At least 300 words are needed for the workspace of a process. ip is the initial priority (optional, may be omitted. default is zero). TRANSFER(VAR p1, p2: PROCESS) Transfer of control between two processes. IOTRANSFER(VAR p1, p2: PROCESS; va: ADDRESS) Transfer procedure for processes operating on a peripheral device. va is an interrupt vector address (not the vector number). LISTEN Procedure to allow an interrupt. The priority is set to 0 and afterwards back to the current priority. SYSRESET Procedure to initialise the system. CODE(c: CARDINAL) Procedure to allow the use of assembly code. c must be constant. SETREG (r: CARDINAL; v: ....); Procedure to set a register to a 32-bit value. r = constant (0..7 = D0..D7, 8..15 = A0..A7). v may be: a variable or constant (the value is put into the register), or a procedure (the entry address is put into the register). Functions --------- ADR(variable): ADDRESS Storage address of the substituted variable. SIZE(variable): CARDINAL Number of bytes used by the substituted variable in the storage. If the variable is of a record type with variants, then the variant with maximal size is assumed. IF variable is not a dynamic array, then SIZE is compatible to CARDINAL, INTEGER and ADDRESS (like a numerical constant). If variable is a dynamicarray, then SIZE is compatible to CARDINAL only (like a CARDINAL variable). TSIZE(type): CARDINAL TSIZE(type, tag1const, tag2const, ... ): CARDINAL Number of bytes used by a variable of the substituted type in the storage. If the type is a record with variants, then tag constants of the last FieldList (see Modula-2 syntax) may be substituted in their nesting order. If no or not all tag constants are specified, then the remaining variant with maximal size is assumed. TSIZE is compatible to CARDINAL, INTEGER and ADDRESS (like a numerical constant). REGISTER(num: CARDINAL): ADDRESS Content of the specified register. num must be a constant: 0..7 = D0..D7, 8..15 = A0..A7. Register A5 points to the process descriptor record, A6 to the data of the current procedure (dynamic link),and A7 to the top of stack. function ASH (x, k): shift x arithmetically by k bits. x may be CARDINAL, INTEGER, ADDRESS or a set. ASH assumes the type of x. k may be INTEGER or CARDINAL. k>0 shifts left, k<0 shifts right. k is only evaluated MOD 64. BIT(i: CARDINAL OR INTEGER): BITSET; Dynamic bit set function for {i}. i may be variable or constant. (whereas in {i} of standard MODULA, i must be a constant; only a procedure is provided for variable i (INCL)). 4.1.8A Workfiles ---------------- The compiler and the linker require workfiles, which can be permanently on the disk. With a fixed disk, or with small programs this is recommended. They should be created (MAKE) by the user when installing Modula-2. Otherwise the compiler and linker will create (and delete) them as required. Debugging needs a dump file, where a memory dump can be written. This file must be created explicitely (MAKE) by you when debugging is desired. File names and maximal file sizes: IL1.WORK 70 KB IL2.WORK 50 KB ASCII.WORK 9 KB LINKER.WORK 10 KB DEBUG.COR memory size + 0.25 KB 4.1.8B Command File Execution ----------------------------- On the SMAKY 8, command files are handled by the executive X, by typing a file name with @, e.g. X @ORDERS The file ORDERS.COM would then contain a list of programs to be run, intermixed with the terminal input those programs expect. Example for contents of ORDERS.COM (compile, link, execute ANYPROG): MODULA ANYPROG.MOD LINKER ANYPROG.LNK ANYPROG <eof> 4.1.8C History (Acknowledgements) --------------------------------- This implementation is a combination and translation of the compilers written by Leo Geissmann (PDP 11 Modula-2 compiler) and Hermann Seiler (MC68000 code generation) . 4.1.9 Compiler error messages ----------------------------- 0 : illegal character 1 : eof in E+ option 2 : constant out of range 3 : open comment at end of file 4 : string terminator not in this line 5 : too many errors 6 : string too long 7 : too many identifiers (identifier table full) 8 : too many identifiers (hash table full) 20 : identifier expected 21 : INTEGER constant expected 22 : ']' expected 23 : ';' expected 24 : block name at the END does not match 25 : error in block 26 : ':=' expected 27 : error in expression 28 : THEN expected 29 : error in LOOP statement 30 : constant must not bee CARDINAL 31 : error in REPEAT statement 32 : UNTIL expected 33 : error in WHILE statement 34 : DO expected 35 : error in CASE statement 36 : OF expected 37 : ':' expected 38 : BEGIN expected 39 : error in WITH statement 40 : END expected 41 : ')' expected 42 : error in constant 43 : '=' expected 44 : error in TYPE declaration 45 : '(' expected 46 : MODULE expected 47 : QUALIFIED expected 48 : error in factor 49 : error in simple type 50 : ',' expected 51 : error in formal type 52 : error in statement sequence 53 : '.' expected 54 : export at global level not allowed 55 : body in definition module not allowed 56 : TO expected 57 : nested module in definition module not allowed 58 : '}' expected 59 : '..' expected 60 : error in FOR statement 61 : IMPORT expected 70 : identifier specified twice in importlist 71 : identifier not exported from qualifying module 72 : identifier declared twice 73 : identifier not declared 74 : type not declared 75 : identifier already declared in module environment 76 : dynamic array must not be value parameter 77 : too many nesting levels 78 : value of absolute address must be of type CARDINAL 79 : scope table overflow in compiler 80 : illegal priority 81 : definition module belonging to implementation module not found 82 : structure not allowed for implementation or hidden type 83 : procedure implementation different from definition 84 : not all defined procedures or hidden types implemented 86 : incompatible versions of symbolic modules 88 : function type is not scalar or basic type 90 : pointer-referenced type not declared 91 : tagfieldtype expected 92 : incompatible type of variant-constant 93 : constant used twice 94 : arithmetic error in evaluation of constant expression 95 : range not correct 96 : range only with scalar types 97 : type-incompatible constructor element 98 : element value out of bounds 99 : set-type identifier expected 101 : undeclared identifier in export-list of the module 103 : wrong class of identifier 104 : no such module name found 105 : module name expected 106 : scalar type expected 107 : set too large 108 : type must not be INTEGER or CARDINAL or ADDRESS 109 : scalar or subrange type expected 110 : variant value out of bounds 111 : illegal export from program module 120 : incompatible types in conversion 121 : this type is not expected 122 : variable expected 123 : incorrect constant 124 : no procedure found for substitution 125 : unsatisfying parameters of substituted procedure 126 : set constant out of range 127 : error in standard procedure parameters 128 : type incompatibility 129 : type identifier expected 130 : type impossible to index 131 : field not belonging to a record variable 132 : too many parameters 134 : reference not to a variable 135 : illegal parameter substitution 136 : constant expected 137 : expected parameters 138 : BOOLEAN type expected 139 : scalar types expected 140 : operation with incompatible type 141 : only global procedure or function allowed in expression 142 : incompatible element type 143 : type incompatible operands 144 : no selectors allowed for procedures 145 : only function call allowed in expression 146 : arrow not belonging to a pointer variable 147 : standard function or procedure must not be assigned 148 : constant not allowed as variant 149 : SET type expected 150 : illegal substitution to WORD parameter 151 : EXIT only in LOOP 152 : RETURN only in procedure or function 153 : expression expected 154 : expression not allowed 155 : type of function expected 156 : INTEGER constant expected 157 : procedure call expected (or ':=' misspelled) 158 : identifier not exported from qualifying module 161 : call of procedure with lower priority not allowed 198 : CARDINAL constant expected 199 : BITSET type expected 200 : size of structured type too large for this processor 201 : array index too large for this element type 202 : array element size too large for this processor 203 : array index type too large for this processor 204 : subrange too large for this processor 206 : illegal subrange type 207 : case label range too large, use IF statement 208 : global data too large for this processor 209 : local data too large for this processor 210 : parameter data too large for this processor 211 : offset of record field too large for this processor 225 : too many unnamed types (typetable full) 226 : reference file too long 300 : index out of range 301 : division by zero 303 : CASE label defined twice 304 : this constant is not allowed as case label 400 : expression too complicated (register overflow) 401 : expression too complicated (codetable overflow) 402 : expression too complicated (branch too long) 403 : expression too complicated (jumptable overflow) 404 : too many globals, externals and calls 405 : procedure or module body too long (codetable) 410 : (UNIX only) no priority specification allowed 411 : (UNIX only) global variable expected 923 : standard procedure or function not implemented 924 : parameter must not be accessed with a WITH 940 : compiler error, adr. cat. violation in PASS4 941 : displacement overflow in ari addressing mode 942 : 32bit by 32bit multiply/divide not yet implemented 943 : index range must not exceed positive integer range 944 : jump too long (overflow in pc-relative offset) 945 : offset too long (overflow in pc-relative offset) 946 : FOR control variable is not of simple addressing mode 974 : step 0 in FOR statement 981 : constant out of legal range 982 : overflow / underflow in index / offset / address calculation 983 : use UNIXCALL instead of standard procedure call 990 : too many WITH nested 991 : CARDINAL divisor too large ( > 8000H ) 992 : FOR control variable must not have byte size (for step # -1, 1) 993 : INC, DEC not implemented with 2. argument for byte variable 994 : too many nested procedures 995 : FOR step too large ( > 7FFFH ) 996 : CASE label too large ( > 7FFFH ) 997 : type transfer function not implemented 998 : FOR limit too large 999 : missing symbol file(s)