JPS6391741A - Table access system - Google Patents

Abstract

PURPOSE:To facilitate the statement of a table and to enhance the maintenability of the table and ability to access by introducing the idea about the branch of programming the syntax of a high-level language into the statement of the table. CONSTITUTION:A machine instruction code (stage) corresponding to one intermediate code is sandwiched with a STAGE statement 4 and an ENDSTAGE statement 5. A CODE I statement 6 through a CODE VIII statement 13 show the generation of machine instructions, which are processed from top to bottom. A SEL statement 14, an OTHER statement 15 and an ENDSEL statement 16 are branch statements corresponding to IF-ELSE-ENDIF. When conditions 14a written in the SEL statement 14 is satisfied, the operation is handed to a CODE III statement 8. Next to the processing of a CODE IV statement 9, a CODE VII statement 12 is processed. If the conditions 14a is not satisfied, the operation is handed to a CODE V statement 10. Following the processing of a CODE VI statement 11, a CODE VII statement 12 is processed. Hence statements can be stated in the table in a lump at every condition.

Description

[Detailed Description of the Invention] [Field of Industrial Application] The present invention relates to a table access method, and in particular, in a compiler, etc., it is possible to determine whether or not each entry in a code table containing a machine instruction code corresponding to an input intermediate code can be accessed. This invention relates to a table access method suitable for determination.

[Conventional technology]

When developing the compiler's machine instruction code generator,
When designing a code table for generating the corresponding machine instruction code from the input intermediate code, it is necessary to make the table description easy, make the generation conditions of the machine instruction code to be generated visible, and make efficient condition judgments. be.

Conventionally, a table is generally described for the entries to be processed, and a determination table or the like is used to determine in advance whether or not each entry in the table can be accessed before performing table access processing. The results are stored in a separate information table, and each entry is accessed while referring to the information in this information table, and no consideration is given to efficient condition determination.

[Problem that the invention seeks to solve]

In the above-mentioned conventional technology, it is not clear in the table description whether each entry can be accessed or not, or the conditions under which a certain entry can be accessed. Furthermore, since the determination of whether or not an entry is accessible and the processing for accessing the entry are separated, maintainability is poor. moreover,
Table access is sequential, and it is always determined whether each entry can be accessed. Therefore, regardless of the entry being accessed, the dynamic step of one determination process is determined by the total number of entries that make up the table. If the number of required entries is small, the processing speed relative to that number of entries will be slow.

SUMMARY OF THE INVENTION An object of the present invention is to improve the problems of the prior art as described above, facilitate table description, improve table maintainability, and improve table access ability.

[Means for solving problems]

The present invention introduces the following four methods.

■ Introduce a description method that uses the concept of branching in well-structured programming in high-level languages to describe tables.

■ The table generation process uses the table descriptions described using the description method described in ■ above, and creates entries corresponding to each description and multiple entries that control access processing by branching, corresponding to the branch descriptions. A table consisting of three elements: the head of an entry group (jump destination), a branching entry consisting of an information storage area and an access image distortion determination condition information storage area, the head (jump destination) of the plurality of entry groups described above, and 7 dresses. generate.

■ The table generation process described in (■) above connects the jump information in the branch entry and the jump destination address on the table during table generation, so each branch and its jump destination are indexed and managed. .

■ Through the table generation processing described in ■ and ■ above, each entry in the generated table is accessed sequentially up to one branch entry, and when processing a branch entry, the corresponding entry Using the judgment condition information in the middle, it is judged whether access is possible for multiple end relays selected by the branch entry being processed, and according to the result, the head of the corresponding entry group, that is, the table of ■ and ■ above, is determined. In the generation process, the process is moved to the corresponding resolved destination address.

(Operation) The table generation process creates a branch entry with a jump destination address, jump light information, and judgment condition information for each branch using the branch description in the table description written using the description method described in (■) above. After indexing the description, the branch entry and the corresponding jump address are generated while being managed.For example, when writing a table using assembler macro instructions, By using a globally variable symbol as an index and combining it with a character string common to branch descriptions, a unique name is generated for a branch, and that name is used to indicate the symbolic name of the jump destination and the jump destination. This is achieved by using an A-type address constant.

In accessing the above table, each entry is sequentially accessed up to the branch entry, and in the branch entry, judgment condition information is used to determine the group of entries selected by the branch entry being processed. It is determined whether or not access is possible, and according to the result, processing is passed to the destination address obtained from the flight information, that is, to the head of the corresponding entry group. For example, when using the above-mentioned assembler macro instruction, if each entry is accessed using a pointer, etc., the result of one condition judgment is the corresponding entry group, that is, the type A address constant indicating the jump destination address. This method is achieved by loading the value into the above pointer.

〔Example〕

An embodiment of the present invention will be described below with reference to FIGS. 1, 2, and 3.

FIG. 1 shows a process for converting intermediate code obtained by branching and rearranging high-level languages during the compilation process into machine instruction code in the machine instruction code generation section of a compiler in an electronic computer system using the present invention. This is an example of code table description, and shows that code table 3 is generated using table generation process 2 from table 1 described in intermediate code.

Machine instruction codes corresponding to one intermediate code are grouped together as one unit, which is called a stage. The stage is sandwiched between 5TAGE statement 4 and ENDSTAGE statement 5. From C0DE 1 sentence 6, C0DE ■ sentence 13 is,
Shows machine instruction generation, processed from top to bottom. SEL
Sentence 14.0 THER Sentence 15. ENDSEL statements 16 are branch statements corresponding to IP-ELSE-ENDIF, and if condition 14a described in SEL statement 14 is satisfied, processing passes to CODEm statement 8, and after processing CoDEIV statement 9, C0DE ■Perform statement 12 and condition 14a
If not satisfied, processing passes to C0DEV statement 10, and C0D
This indicates that the C0DE ■ statement 12 is processed next to the EVI statement 11. Therefore, tables can be written for each condition, and just by looking at the written table 1,
It can be seen what kind of machine instruction code will be generated.

Using table generation processing 2 from the above table description statement, 5TAGE statement 4 and 5TAGE from stage name 4a.
An entry 17 and a stage head label 29 are generated. Also, from ENDSTAGE statement 5, ENDSTAG
E entry 18 is generated. Also, C0DE 1 sentence 6
From the C0DE ■ statement 13, machine instruction generation entries 19 to 26 are generated. Also, from the SEL statement 14, the SEL entry 27 that holds the address 30a of A30, the address 31a of B31, the comparison operator information 32, and the address 33 of the jump destination label JQa34 when the condition is not met is used for the condition 14a. is generated. Also, 0THE
From the R statement 15, "hold the address 35 of the jump destination label JQb36" after the processing when the condition of the corresponding SEL statement 14 is satisfied.

0THER entry 28 and label Qa 34 are generated. Furthermore, a label Qb36 is generated from the ENDSEL statement 16.

2(A) and 2(B) show the access processing procedure for the code table 3 generated in FIG. 1, and FIG. 3 shows the code table 3 generated in FIG. 1 and the pointer used for accessing the code table 3. Memory in 53 electronic computers 5
2 shows how the code table 3 is accessed by the pointer 53 at each point in the process.

When control is passed to the machine instruction code generation section and machine instruction generation processing is started, an intermediate code is input from the intermediate code file 39 (step 38), and the end of the file is detected by determining the end of the file (step 40). Repeat the following process and the above input until the When the process is finished, the machine instruction generation process ends (step 41).

According to the intermediate code information input from the intermediate code file 39, the value of the pointer 53 is set to "the beginning of the corresponding stage".
29 (step 42), points to 54a in FIG. 3, and repeats the following process until the process returns to input 38.

First, it is determined whether the entry is SEL entry 27 (step 43). If SEL entry 27 (54c) -
A36 (7) determines whether the condition is satisfied or not (step 50).
) 7 dresses 30. This is performed using the B57(7)7 address 31 and comparison operator information 32.

In this condition determination 50, the condition is established (A = B
), change the value of pointer 53 to "next entry" (step 49), point to 54d in FIG.
If the condition is not satisfied (A to B), change the value of the pointer 53 to "Jump destination label JQa34 using address 33 of jump destination label JMa34 (step 51), and point to 44f in FIG. .On the other hand, SEL entry 2
In the determination 43 of 7, if the SEL entry 27 is not 7 (54a, 54h, 54i in FIG. 3), 0THER
It is determined whether the entry is entry 28 (step 44). 0
If the THER entry is 28 (54e), change the value of the pointer 53 to "Jump destination label JQb36" using address 35 of "Jump destination label" 12b36 (Step 51), and point to 54g.0 THER entry If it is not 28 (54a, 54h, 54i),
The ENDSTAGE engine 18 makes a determination (step 45). ENDSTAGE entry 1
If it is not 8 (54a, 54h), perform machine instruction generation processing 46 corresponding to each entry, output the generated machine instruction code to the target program file 48 (step 47), and change the value of pointer 53 to ``next entry''. ” (step 49).

Point to 54bi. E N D S T A G
If the E entry is 18 (54i), the process returns to input 38.

According to this embodiment, the written Table 1 can be easily written because conditions can be judged at any location, and it is possible to visually recognize what kind of machine instruction will be generated in the machine instruction code generation process. This improves the maintainability of the machine instruction code generation process, allows the machine instruction codes corresponding to one intermediate code to be written as one, and allows the machine instruction code to be written without using decision tables. Trial speed is improved because the generation conditions can be determined.

〔Effect of the invention〕

As explained above, according to the present invention, the description of a table is made easier by introducing the concept of branching in well-structured programming in a high-level language, and the conditions for whether or not each entry can be accessed are set in the table description. Since it is visually clear, maintainability of the table is improved, and since access processing is performed only when access is required for each entry in the table, it has the effect of improving access ability.

[Brief explanation of the drawing]

FIG. 1 is a diagram showing the table description of a machine instruction code generation code table in a compiler according to an embodiment of the present invention, the table generation process, and the relationship between the code table generated by the table generation process. Figure 3 shows the flow of processing control for the code table access method shown in Figure 1. FIG. 3 is a diagram showing the position on the table pointed to by the pointer. 1... Table description example, 2... Table generation process, 3... Generated code table. Figure 2 (3)

Claims (1)

[Claims] (1) For tables consisting of descriptions of entries that store information that constitutes multiple table components, the concept of branching in well-structured programming in high-level languages is introduced, and branches are provided to select entries to be processed. A table access method characterized in that an entry to be processed is selected by the branch described above when the described table is accessed.