JPH01177128A - Assembler system - Google Patents

Abstract

PURPOSE:To decrease the linking jobs of an post-process and to increase the assembling speed by decreasing the number of pending symbols to be processed in the post-process. CONSTITUTION:An assembler of each module extracts a character string and its value of a local symbol defined in a module as well as a character string of a public symbol defined in another module that is referred to in the relevant module and stores these extracted character strings and values into a symbol table 57a set in a main memory 57. Then the character string and its value of a public symbol defined in the relevant module are extracted and stored in a public symbol table 56b set in an auxiliary memory 56. At the same time, the value of the public symbol whose character string only is stored in the table 56b is searched out of the table 56b for substitution. As a result, the number of pending symbols can be decreased and a high-speed assembling job is attained.

Description

[Detailed Description of the Invention] [Field of Industrial Application] The present invention relates to an assembler method for converting a source program written in assembly fairy tale, which is a programming language for microcomputers, into machine language, and in particular, the present invention relates to an assembler method for converting a source program written in assembly fairy tale, which is a program description language for microcomputers, into machine language. This book concerns the assembler system for programs in which symbols are referenced between these modules.

[Prior Art] Generally, an assembler is a program for converting a source program written in assembly language into machine language so that it can be directly executed by a microcomputer. The operating environment of the assembler is, for example, as shown in FIG.
K) and a central processing unit 51 that assembles the source program loaded from this auxiliary memory bag W50.
(CPU), a main memory device 52 (MM) used for the assembling process, an operation system 53 (O3) that controls the execution of the central processing unit 51, and a system for inputting commands to the operation system 53. It also includes a keyboard 54 and a display 55 for outputting processing results.

Execution of the assembler is initiated by entering commands into the operating system 53 from the keyboard 54. First, an assembler program is loaded from the auxiliary storage device 50 to the main storage device 52, and this program is executed by the central processing unit 51. According to execution of the instructions of the assembler program, the source program is sequentially read from the auxiliary storage device 50 to the main storage device 52 and converted into machine language. The output machine language is stored in the auxiliary storage device 50 as an object file.

By the way, two assembler systems are conventionally known: a so-called relocatable assembler and a so-called absolute assembler.

These two methods will be explained below.

A relocatable assembler is an assembler method that outputs machine language that can be relocated to any address in memory, and has the advantage that programs can be easily modified. In this method, the source
The program 61 is divided into multiple modules 1,
In an input step 62, these modules 1, 2.3 are input individually from the keyboard 54 in FIG. The relocatable assembler 64 separately assembles these source module files 63 and converts the machine language object program into a relocatable object file 6.
Output in format 5. Relocatable assembler 64
converts each source module file 63 into machine language completely independently, so the output locator
The memory address assigned to each instruction of the machine language program in the object file 65 is a relative address so that it can be determined independently for each module. In order to make the machine language program of this relocatable object file 65 into a format that can be executed by a computer,
Furthermore, it is necessary to perform a linking operation using the linker 66 and change the relative address in the relocatable object file 65 to an absolute address in the memory. The machine language program reassigned to the absolute address by the linker 66 is stored as a load module file 67.

With this relocatable assembler method, there is no need for the program to be aware of the order in which modules 1, 2, and 3 are organized; in other words, the order of the modules only needs to be specified at the time of linking. Sometimes, if other module or locator object files already exist, you only need to assemble the modified module, and then simply link the currently assembled or locator object with the other module or locator object files. This means that a modified load module will be generated just by doing this. This feature has the advantage of reducing assembly time when a large program requires minor modifications, since fewer source modules need to be assembled. However, since the relocatable assembler method assembles each module independently, somewhat complicated management is required, for example, when a symbol referenced in one module is defined in another module.

That is, assembly language has pseudo-instructions that name arbitrary data or addresses. When a name to which a certain data or address value is assigned by the pseudo-instruction is written in the source program, the name is converted to the corresponding value during assembly. This name is generally called a symbol, and writing a symbol in a source program is called symbol reference. When a symbol is referenced, the value assigned to the symbol is assigned to the position where the symbol is written. Since this is necessary, it is necessary to create a symbol table in a part of the main storage device 52 in advance in which the character strings and values of symbols are compared.

For this reason, the assembler usually uses a two-pass assembler method, and the source program is analyzed in two parts. In other words, in pass 1, which is the -th analysis phase, locations in the source program where symbols are defined are extracted, and a symbol table is created in the main memory that associates symbols with their assigned values. Ru.

Then, in pass 2, which is the second analysis phase, the value searched from the symbol table is filled in the symbol reference location.

Even in relocatable assemblers, if symbols are defined and referenced only within one module, the above assembly method is sufficient, but if symbols are defined and referenced between different modules, Special management is required. Generally, the former is
Symbols, the latter are called public symbols, now,
If a module references a public symbol defined in another module, that public symbol will not be registered in the symbol table created in pass 1 of assembling, so pass 2 of assembling will
, its value cannot be referenced. Therefore, it is output to the locator object file 65 without being resolved. That is, for example, as in the source program 70 shown in FIG. 9(a), modules 71, 7
Symbols SYMA, SYMB, and SYMC defined at relative addresses 10, 50.80, respectively, of 2.73 are referenced by the BR instruction in each external module, but modules 71, 72.73 are assembled individually. For,
“BRSMMC” of module 71, “BRS Y M A” of module 72, “BR” of module 73
SYMB” instruction operands are not converted to numerical values,
The data is output to each relocatable object file 65 unresolved. Therefore, the relatively allocated memory addresses of these relocatable object files 65 are converted into absolute addresses 0-590, as shown in FIG. Symbols SYMA and SYMB that were resolved
, SYMC and outputting it as the load module file 74 is performed by the linker 6 in FIG.
6.

On the other hand, in contrast to the relocatable assembler, in the absolute assembler, the memory addresses allocated to instructions and data are absolute addresses.

That is, as shown in FIG. 10, a source file 83 obtained by inputting a source program 81 from the keyboard 54 in FIG. 7 in an input step 82 is converted into an object file 85 by an assembler 84. This object file 85 can be directly loaded into a computer and executed as a load module file. Further, symbol management in the absolute assembler is performed only by the symbol table in the main storage device 52. When referencing a symbol, a method is used in which the symbol table is searched, a value corresponding to the symbol is extracted, and the extracted value is assigned to the symbol description position.

In the absolute assembler method, there is one source module and one object module, which is the load module, so there is no need to refer to an external module for symbols as in the sorlocator assembler, so there are no unresolved symbols in the assembler. etc. do not exist, and since absolute addresses are allocated to memory, the link processing in the relocatable assembler is unnecessary in the absolute assembler.

[Problems to be Solved by the Invention] In the conventional locator assembler described above, symbol tables are created individually when assembling each module that makes up a source program, so one symbol table is small. Symbol search time is shortened and assembly time is also shortened. However, since unresolved symbol references must be resolved by link processing after assembly is completed, there is a drawback that link processing takes time when there are a large number of public symbols.

On the other hand, in the absolute assembler method, where only one symbol table is created for one source program, the symbol table grows as the source program expands, so it takes a lot of time to search for symbols. The disadvantage is that it consumes a lot of time and the assembly speed is slow.

The present invention has been made in view of such problems, and includes:
It is an object of the present invention to provide an assembler system that can shorten symbol search time and link processing time, and can perform high-speed assembly.

[Means for Solving the Problems] The present invention is a divided module type source program in which at least one module includes a public symbol reference instruction, and the source program is assembled into each module in a prespecified organization order. This is an assembler method for generating object files, and is characterized in that the assembler for each module includes the following two procedures.

The first step is to extract the character strings and values of local symbols defined in the module being processed, and the character strings of public symbols referenced within the module and defined in other modules, This is the procedure for storing in the symbol table.

The second step is to extract the character strings and values of public symbols defined in the module and store them in the public symbol table. A value of a public symbol is searched from the public symbol table and assigned.

Note that the symbol table is formed for each module in the main memory. Further, the public symbol table is formed in the auxiliary storage device as one common to all modules.

[Operation] According to the present invention, since the source program is divided into modules, the symbol table in each module can be made smaller, and the symbol search time can be shortened.

Furthermore, in the present invention, when assembling a source program divided into modules, the organization order of the modules is specified in advance, and the character strings of public symbols defined in each module and the values assigned to them are specified at the time of assembly. By creating the associated public symbol table in the auxiliary storage device, even if the symbol table is destroyed when the assembly of one module is completed, the public symbol table remains undestroyed. For this reason, if a public symbol referenced in a module is defined in a module organized earlier, it is not possible to output that symbol unresolved to an object file, as in a relocatable assembler. Instead, the values of the symbols stored in the public symbol table in the auxiliary storage described above are referenced during assembly to create objects.
It can be output to a file. In other words, the number of unresolved symbols sent to post-processing can be reduced.

Furthermore, in the present invention, absolute addresses of machine language instructions can be assigned at the time of assembly by specifying the order of module organization and searching the public symbol table.

Therefore, the post-processing work of the linker can be greatly reduced and the assembly speed can be increased.

[Example] Next, an example of the present invention will be described in detail with reference to the accompanying drawings.

FIG. 1 shows an operating environment for an assembly method according to an embodiment of the present invention. The main difference from the system shown in FIG. 7 is the contents of the auxiliary storage device 56.

In addition to the source program and the improved assembler program, the auxiliary storage device 56 stores a module organization order table 56a that specifies the organization order of modules, and a public symbol store that stores public symbols defined in the source program. - A table 56b is provided. Furthermore, a symbol table 57a is formed in the main storage device 57, which is rewritten for each module.

The assembly method is a two-bus assembly method for each divided module. In pass 1, an EXTRNffJ-like instruction is used that extracts a local symbol from the source module, stores its character string and value in the symbol table 57a, and declares that it refers to a symbol defined in another module. EXTRN defined by
Extract the (external) symbol and convert the string to the symbol
It is stored in table 57. In pass 2, the source
Extracts a public symbol from the module, stores its character string and value in the public symbol table 56b, and stores the unresolved EXTR,N symbols and their corresponding values stored in the symbol table 57a as public symbols. - Find it from the table 56b and output it to the symbol table 57a. When searching for the value of the EXTRN symbol in the public symbol table 56b, the contents of the module organization order table 56a are referred to, and the public symbol extracted in the module organized before the source module currently being assembled is searched for. Only symbols are searched.

Next, the operation of the present invention will be explained in detail with reference to the flowcharts of FIGS. 2 and 3.

FIG. 2 is a flow diagram illustrating symbol-related operations in pass 1 of a two-pass assembler.

In step C1, one line is extracted from the source module. In step C2, it is determined whether or not a symbol definition pseudo-instruction that defines arbitrary data or address value in the name is written in the one line extracted in step C1. to decide. If a symbol definition directive is written,
At step C5, the symbols and assigned values are stored in the symbol table 57a, and the process returns to step C1. If it is not a symbol definition pseudo-instruction, it is determined in step C3 whether it is an EXTRN pseudo-instruction that explicitly declares the symbol to refer to a symbol defined outside the module, and if this EXTRN pseudo-instruction is written, In step C6, the character string of the symbol and a character (for example, E) indicating that it is an EXTRN symbol are stored in the symbol table 57a, and the process returns to step C1. If it is not an EXTRN pseudo-instruction, processing other than symbols is finished in step C4, and the process returns to step C1.

FIG. 3 is a flow diagram showing symbol-related operations in pass 2 of the two-pass assembly method. In pass 2, the source program is interpreted again from the beginning. In step D1, one line is extracted from the source module. In step D2, PUBLI declares the symbol ・PtJBLI C in the line extracted in step D1.
If a C pseudo-instruction is written, the symbol string written in that line is searched for in the symbol table 57a as shown in FIG. 1 in step D6, and the string and value of the found symbol are searched in step D5. It is stored in the public symbol table 56b, and the process returns to step D1.

Next, if it is determined in step D2 that it is not a PUBLIC pseudo-instruction, it is determined in step D3 whether or not a symbol reference is made in that line.

If a symbol is referenced, the symbol table 57a is searched for the same character string as the symbol in step D7, and in step D8, E is written in the column of the found symbol, and the symbol refers to the value defined in the external module. If it is determined that it is, in step D10,
A symbol having the same character string as that symbol is searched from the public symbol table 56b created when assembling the module in the same assemble unit, and step Dl
If it is determined in step D9 that the symbol is stored in the public symbol table 56b, the value assigned to that symbol is substituted in step D9. Further, if it is determined in step D11 that the symbol does not exist in the public symbol table 56b, it is held as an unresolved symbol in step D12.

Furthermore, if E is not written in the symbol field of the symbol table in step D8, the value assigned to the symbol is extracted and substituted into that symbol. When the processing in step D9 or D12 is completed, the process returns to step D1. Note that when a symbol is referenced and the symbol is not described in the symbol table, there are steps such as outputting a symbol undefined error, but the explanation is omitted here.

Taking the source program of FIG. 9(a) described above as an example, the above processing is performed as follows.
, (b).

4, 5, and 6 show the modules 71 and 72, respectively.
The contents of the symbol table and public symbol table for 73 are shown, and in each figure, (a) is the symbol table and (b) is the public symbol table.
Shows the contents of the table.

It is assumed that the module organization order table 56a specifies the organization order in the order of modules 71, 72, and 73.

First, in pass 1 of assembly of module 71, symbol definition pseudo-instruction SYMA is extracted, and its character string SY
MA and the value VALUEA are stored in the symbol table 57a. Next, the EXTRN pseudo-instruction is extracted, and its character string SMMC and the character ``E'' indicating that its value is undefined are stored. [Figure 4(a)].

In pass 2, the PUBLIC pseudo-instruction is extracted, and by searching the symbol table 57a, its character string SYMA
and the value VALUEA are stored in the public symbol table 56b [FIG. 4(b)]. Furthermore, the symbol reference instruction BR is extracted, and its symbol SMMC is EX
Since it is a TRN symbol, the public symbol table 56b is searched, but in this case, the last SYMC
Since the SYMC value is not registered, the process ends with the SYMC value undetermined.

In the assembling process of the module 72, in pass 1, the symbol definition pseudo-instruction SYMB and the EXTRN pseudo-instruction SYM are
A is extracted, and the contents of the symbol table 5'7b become as shown in FIG. 5(a). In pass 2, the PUBLI C pseudo-instruction SYMB is newly registered in the public symbol table 56b. In this pass 2, the symbol reference instruction BR specifies the symbol SYMA. At this time, public symbol table 56
Since the value VALUEA of SYMA is registered in SYMA b [FIG. 5(b)], SY of the symbol table 57a
The above VALUEA can be substituted as the value of MA. Therefore, symbol table 5 of module 72
At 7a, the value of the public symbol SYMA is determined.

Similarly, regarding the module 73, since the value of SYMB of the public symbol referenced by the module 73 is already registered in the public symbol table 56b, the value of SYMB of the symbol table 57a must be determined. can be done [Fig. 6 (a), (b)].

Therefore, when the assembly of the three modules 71, 72, and 73 is completed, the only unresolved symbol is SYMC in module 71, and the burden of subsequent link processing can be greatly reduced.

In this way, in the source program of FIG. 9(a), the symbol SY referenced in modules 72 and 73
Since MA and SYMB are defined in modules with young addresses, the values stored in the symbol table are
It can be assigned to MA and SYMB during assembly.

In the explanation of the above embodiment, the auxiliary storage device used to store public symbols is not only a device with a low access speed such as a so-called disk, but also a so-called RAM disk made of semiconductor memory or a bubble memory device.・It may be composed of memory or the like.

Finally, it will be explained that the assembler of the present invention described above is particularly effective in so-called single-chip microcomputers. In single-chip microcontrollers, the program memory (ROM) and data memory (RAM) in the main memory usually have different address spaces, and because the data storage area is severely constrained, symbol definitions are limited to the source program. There is a custom of doing this at the beginning of the

Undefined symbols are concentrated in labels that indicate program memory addresses, but because of program locality, labels are overwhelmingly local symbols. Therefore, in single-chip microcontrollers, most public symbols are located in modules with lower rear addresses, and as a result, most of them are resolved during assembly, which dramatically reduces the number of unresolved symbols at the end of assembly. That's what I do. Therefore, when the assembler method of the present invention is applied to an assembler that converts a source program for a single-chip microcontroller into machine language, the assembler method of the present invention has a faster assembling speed than a conventional assembler.
It is expected that this will improve significantly.

[Effects of the Invention] As explained above, according to the assembler method according to the present invention, when a source program is divided into multiple parts and modularized, local symbols that can be referenced only within the module in which they are defined is stored in main memory, and in addition, public symbols that are referenced even from external modules other than the module that defined the symbol are stored in auxiliary memory so that they remain even if the symbol table in main memory is destroyed. By storing public symbols in any public symbol table in the storage device, public symbols defined in modules that are organized earlier in assembly order can be searched and referenced in the public symbol table, and converted into machine language. can be converted. Therefore, after the assembly is completed, it is possible to reduce the number of so-called unresolved symbols whose memory addresses to be loaded have been determined but whose values have not been assigned. Furthermore, since the order of organization of modules at the time of assembly is specified, machine language instructions can be assigned to absolute addresses at the time of assembly. Therefore, the work of the linker can be greatly reduced and the assembly speed can be significantly increased, which is a great advantage.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of a system that executes an assembly method according to an embodiment of the present invention, FIG. 2 is an operation flow diagram of pass 1 of the embodiment, and FIG. 3 is an operation flow diagram of pass 2 of the embodiment. , FIGS. 4(a), (b), 5[](a), (b), and FIGS. 6(a), (b) are the modules 71, 72., respectively.
7 is a schematic diagram showing the contents of a symbol table and a public symbol table in 73, FIG. 7 is a block diagram showing a schematic configuration of a conventional assemble system, and FIG.
The figure is a block diagram showing the schematic configuration of a conventional glue locator pull assembler, and FIGS.
A schematic diagram of a module file. FIG. 10 is a block diagram showing a schematic configuration of a conventional absolute assembler. 50.56: Auxiliary storage device, 51; Central processing unit, 52
．． 57. Main storage device, 53; operation system,
54; keyboard; 55; display; 56a; module organization order table; 56b; public symbol table; 57a; symbol table

Claims (1)

[Claims] A source program that is written in assembly language and is divided into a plurality of modules, and in which at least one module includes instructions for referencing public symbols defined in other modules, is divided into a plurality of modules in a pre-specified organization order for each of the modules. This is an assembler method that generates an object file by assembling into objects, and the assembler for each module assembles the character strings and values of local symbols defined in the module, and the strings and values of local symbols that are referenced in the module and in other modules. public defined in
a first step of extracting character strings of symbols and storing them in a symbol table formed in main memory; extracting character strings and values of public symbols defined in the module; These are stored in a public symbol table formed in an auxiliary storage device, and the value of the public symbol in which only character strings are stored in the symbol table is searched from the public symbol table and assigned. An assembler method characterized by comprising a second procedure.