asmlnk.doc

很少见的源码公开的msc51和z80的c编译器。

           The  ASxxxx  assemblers  were  written following the style of
        several cross assemblers found in the Digital Equipment Corpora-
        tion  Users  Society  (DECUS)  distribution of the C programming
        language.  The DECUS code was provided with no documentation  as
        to  the  input  syntax  or the output format.  Study of the code
        revealed that the unknown author of the code  had  attempted  to
        formulate  an assembler with attributes similiar to those of the
        PDP-11 MACRO assembler (without macro's).  The  incomplete  code
        from  the  DECUS C distribution has been largely rewritten, only
        the program structure, and C source  file  organization  remains
        relatively  unchanged.   However, I wish to thank the author for
        his contribution to this set of assemblers.  

           The  ASLINK  program was written as a companion to the ASxxxx
        assemblers, its design and implementation was not  derived  from
        any other work.  

           The  ASxxxx  assemblers  and the ASLINK relocating linker are
        placed in the Public Domain.   Publication  or  distribution  of
        these programs for non-commercial use is hereby granted with the
        stipulation that the  copyright  notice  be  included  with  all
        copies.  

           I  would  greatly  appreciate  receiving  the  details of any
        changes, additions, or errors pertaining to these  programs  and
        will  attempt  to  incorporate  any  fixes  or  generally useful
        changes in a future update to these programs.  



                Alan R.  Baldwin 
                Kent State University 
                Physics Department 
                Kent, Ohio 44242 


                [e-mail] 
                tel:  [phone] 
                fax:  [phone] 














                                    CHAPTER 1

                                  THE ASSEMBLER





        1.1  THE ASXXXX ASSEMBLERS 


           The  ASxxxx  assemblers are a series of microprocessor assem-
        blers written in the C programming language.  Each assembler has
        a device specific section which includes:  

             1.  device  description, byte order, and file extension in-
                 formation 

             2.  a  table  of  the assembler general directives, special
                 device directives, assembler mnemonics  and  associated
                 operation codes 

             3.  machine specific code for processing the device mnemon-
                 ics, addressing modes, and special directives 

        The device specific information is detailed in the appendices.  

           The assemblers have a common device independent section which
        handles the details of file input/output, symbol  table  genera-
        tion,  program/data  areas,  expression  analysis, and assembler
        directive processing.  

        The assemblers provide the following features:  

             1.  Command string control of assembly functions 

             2.  Alphabetized, formatted symbol table listing 

             3.  Relocatable object modules 

             4.  Global symbols for linking object modules 

             5.  Conditional assembly directives 



        THE ASSEMBLER                                           PAGE 1-2
        THE ASXXXX ASSEMBLERS


             6.  Program sectioning directives 


           ASxxxx assembles one or more source files into a single relo-
        catable ascii object file.  The output of the ASxxxx  assemblers
        consists of an ascii relocatable object file(*.rel), an assembly
        listing file(*.lst), and a symbol file(*.sym).  


        1.1.1  Assembly Pass 1 


           During  pass  1, ASxxxx opens all source files and performs a
        rudimenatry assembly of each source statement.  During this pro-
        cess  all symbol tables are built, program sections defined, and
        number of bytes for each assembled source line is estimated.  

           At the end of pass 1 all undefined symbols may be made global
        (external) using the ASxxxx switch -g, otherwise undefined  sym-
        bols will be flagged as errors during succeeding passes.  


        1.1.2  Assembly Pass 2 


           During  pass  2  the ASxxxx assembler resolves forward refer-
        ences and determines the number  of  bytes  for  each  assembled
        line.   The  number  of bytes used by a particular assembler in-
        struction may depend upon the addressing mode, whether  the  in-
        struction allows multiple forms based upon the relative distance
        to the addressed location, or other factors.   Pass  2  resolves
        these cases and determines the address of all symbols.  


        1.1.3  Assembly Pass 3 


           Pass 3 by the assembler generates the listing file, the relo-
        catable output file, and the symbol tables.  Also during pass  3
        the errors will be reported.  

           The  relocatable object file is an ascii file containing sym-
        bol references and definitions, program  area  definitions,  and
        the  relocatable assembled code, the linker ASLINK will use this
        information to generate an absolute load file (Motorola or Intel
        formats).  




        THE ASSEMBLER                                           PAGE 1-3
        SOURCE PROGRAM FORMAT


        1.2  SOURCE PROGRAM FORMAT 



        1.2.1  Statement Format 


           A source program is composed of assembly-language statements.
        Each statement must be completed on one line.  A line  may  con-
        tain a maximum of 128 characters, longer lines are truncated and
        lost.  

           An  ASxxxx  assembler  statement  may  have  as  many as four
        fields.  These fields are identified by their order  within  the
        statement  and/or  by separating characters between fields.  The
        general format of the ASxxxx statement is:  

              [label:]  Operator        Operand         [;Comment(s)] 

           The  label and comment fields are optional.  The operator and
        operand fields are interdependent.  The operator field may be an
        assembler  directive or an assembly mnemonic.  The operand field
        may be optional or required as defined in  the  context  of  the
        operator.  

           ASxxxx  interprets  and  processes source statements one at a
        time.  Each statement causes a particular operation to  be  per-
        formed.  


        1.2.1.1  Label Field  - 

           A  label is a user-defined symbol which is assigned the value
        of the current location counter and entered into  the  user  de-
        fined  symbol  table.   The  current location counter is used by
        ASxxxx to assign memory addresses to the source  program  state-
        ments as they are encountered during the assembly process.  Thus
        a label is a means  of  symbolically  referring  to  a  specific
        statement.  

           When  a program section is absolute, the value of the current
        location counter is absolute;  its value references an  absolute
        memory  address.   Similarly, when a program section is relocat-
        able, the value of the current location counter is  relocatable.
        A  relocation  bias  calculated at link time is added to the ap-
        parent value of the current location counter  to  establish  its
        effective  absolute  address  at  execution time.  (The user can
        also force the linker to relocate sections defined as  absolute.
        This may be required under special circumstances.) 

           If  present,  a  label  must  be  the first field in a source
        statement and must be terminated by a colon (:).   For  example,


        THE ASSEMBLER                                           PAGE 1-4
        SOURCE PROGRAM FORMAT


        if  the  value  of  the  current  location  counter  is absolute
        01F0(H), the statement:  

              abcd:     nop 

        assigns  the  value  01F0(H) to the label abcd.  If the location
        counter value were relocatable, the final value of abcd would be
        01F0(H)+K, where K represents the relocation bias of the program
        section, as calculated by the linker at link time.  

           More  than  one label may appear within a single label field.
        Each label so specified is assigned the same address value.  For
        example,  if  the  value  of  the  current  location  counter is
        1FF0(H), the multiple labels in the following statement are each
        assigned the value 1FF0(H):  

              abcd:     aq:     $abc:   nop 

           Multiple labels may also appear on successive lines.  For ex-
        ample, the statements 

              abcd:  
              aq:  
              $abc:     nop 

        likewise  cause  the  same value to be assigned to all three la-
        bels.  

           A  double  colon  (::)  defines the label as a global symbol.
        For example, the statement 

              abcd::    nop 

        establishes the label abcd as a global symbol.  The distinguish-
        ing attribute of a global symbol is that it  can  be  referenced
        from  within an object module other than the module in which the
        symbol is defined.  References to this label  in  other  modules
        are  resolved when the modules are linked as a composite execut-
        able image.  

        The legal characters for defining labels are:  

                A through Z 
                a through z 
                0 through 9 
                . (Period) 
                $ (Dollar sign) 
                _ (underscore) 

           A  label may be any length, however, only the first eight (8)
        characters are significant and, therefore must be  unique  among
        all   labels  in  the  source  program  (not  necessarily  among


        THE ASSEMBLER                                           PAGE 1-5
        SOURCE PROGRAM FORMAT


        separately compiled modules).  An error code(s) (m or p) will be
        generated  in the assembly listing if the first eight characters
        in two or more labels are the same.  The m code is caused by the
        redeclaration  of  the symbol or its reference by another state-
        ment.  The p code is generated because the symbols  location  is
        changing on each pass through the source file.  

           The  label  must  not  start with the characters 0-9, as this
        designates a local symbol with special attributes described in a
        later section.  

           The  label  must  not  start  with  the  sequence $$, as this
        represents the temporary radix 16 for constants.  


        1.2.1.2  Operator Field  - 

           The  operator field specifies the action to be performed.  It
        may consist of an instruction mnemonic (op code) or an assembler
        directive.  

           When  the  operator is an instruction mnemonic, a machine in-
        struction is generated and the assembler evaluates the addresses
        of  the operands which follow.  When the operator is a directive
        ASxxxx performs certain control actions or processing operations
        during assembly of the source program.  

           Leading  and  trailing  spaces  or tabs in the operator field
        have no significance;  such characters serve  only  to  separate
        the operator field from the preceeding and following fields.  

           An operator is terminated by a space, tab or end of line.  


        1.2.1.3  Operand Field  - 

           When  the  operator is an instruction mnemonic (op code), the
        operand  field  contains  program  variables  that  are  to   be
        evaluated/manipulated by the operator.  

           Operands  may  be  expressions  or  symbols, depending on the
        operator.  Multiple expressions used in the operand  fields  may
        be  separated  by a comma.  An operand should be preceeded by an
        operator field;  if it is not, the statement will give an  error
        (q  or  o).   All  operands  following instruction mnemonics are
        treated as expressions.  

           The operand field is terminated by a semicolon when the field