JP3915208B2 - Compilation device - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a compiler that compiles an input source program to generate an object program, and particularly relates to initialization of a dynamic structure variable.
[0002]
[Prior art]
A computer program is usually an executable program by creating a source program (source code) described in some programming language, converting it into object code by a compiler, and linking one or more object codes.
[0003]
Many programming languages support structure variables as their grammar conventions. In addition, there are two types of variables: static variables whose storage area is acquired when an executable program is created, and dynamic variables whose storage area is dynamically acquired when the program is executed. .
Conventional compilers can set initial values for static variables (specify initial values when declaring variables), but cannot set initial values for dynamic structure variables. It was possible. This is because a static variable acquires a data area at the time of compilation, so it is possible to set a value at the time of compilation, whereas a dynamic variable has a data area at the time of compilation. This is because the value cannot be set during compilation.
[0004]
[Problems to be solved by the invention]
For this reason, the program developer had to describe setting of values for the variables separately from the declaration of the variables when creating the source program. In addition, since the value setting is described as a value setting for each element constituting the structure variable, the size of the executable program has been enlarged.
[0005]
Also, it was not possible to check whether or not a value has been set for each element of the structure when referring to the element.
The present invention solves the problem that an initial value cannot be set for a dynamic structure variable and other problems that occur in relation to this problem.
[0006]
[Means for solving problems]
Therefore, in the present invention, a compile apparatus that inputs a source program and outputs an object program, and an initial value setting area for storing an initial value to be set in a dynamic variable is static in the output object program. Initial value setting area acquisition means for acquiring, initial value setting area initialization means for setting an initial value for the acquired initial value setting area, and dynamic variable area dynamically acquired at the time of program execution In this case, the above problem is solved by a compiling device having variable initialization object code generation means for generating code for copying the contents of the initial value setting area to the dynamically acquired area.
[0007]
DETAILED DESCRIPTION OF THE INVENTION
FIG. 1 is a diagram schematically showing how a program is executed on a computer. The computer 4 includes a CPU (not shown), a main storage device (memory), an external storage device, a display device, a keyboard, and the like. Programs and data are stored as files in the external storage device. The execution format program 1 stored in the file of the external storage device is loaded onto the memory and executed by the CPU. The OS 3 controls computer hardware resources and executable programs. The library 2 is a kind of executable program, and is commonly called from a plurality of executable programs.
[0008]
In order to create the executable program 1, an editor 5, a compiler 6 and a linker 7 shown in FIG. 2 are used. The editor 5, the compiler 6, and the linker 7 are all executable programs. The editor 5 is used to create a source program 8 that is the basis of the executable program. The created source program is stored in the file as a source file.
[0009]
The compiler 6 is a program for converting the source program 8 into the object program 12, and includes a syntax analysis unit 9, an optimization processing unit 10, and a code generation unit 11, as shown in FIG.
The syntax analysis unit 9 reads the source program 8 stored in the file, performs syntax analysis according to a predetermined grammar rule, and generates a corresponding intermediate language. The optimization processing unit 10 converts the intermediate language generated by the syntax analysis unit 9 so that the execution efficiency is improved. The code generation unit 11 converts the intermediate language obtained by conversion by the optimization processing unit 10 into an object program 12 suitable for a computer on which the program is executed.
[0010]
As shown in FIG. 4, a plurality of object programs (12-1, 12-2, 12-3, etc.) generated by the compiler 6 are combined by a linker 7 to generate an executable program 1.
The compiler 6 is activated by inputting a command for executing the compiler 6 from the keyboard. When the compiler 6 is started, a command is used together with commands to specify the name of the source file storing the source program and various instructions to the compiler. An example of the option is an undefined value detection instruction option at the time of execution.
[0011]
The undefined value detection instruction option checks whether or not a value has already been assigned to a variable that is used in the program during program execution. This is an option for creating an executable program that displays an error message if there is not. When this option is instructed, it is necessary to perform some specific processing prior to referring to the variable. Therefore, the execution efficiency of the program is lowered, but it is very useful for debugging the program.
[0012]
When the syntax analysis unit 9 recognizes the definition of the dynamic structure variable during the syntax analysis of the source program, the processing shown in FIG. 5 is performed.
First, in step S21, it is checked whether an undefined value detection option is instructed. If an option is instructed, the process proceeds to step S22. If not, the process proceeds to step S27.
[0013]
When the option is instructed, a template area used for initializing the dynamic structure pair variable is secured in step S22. Here, when a template for a dynamic structure variable of the same type is already secured, a new template area is not secured. The template area is finally secured in a static data area included in the execution format program 1.
[0014]
Next, in steps S23, S24, S25, and S26, a value is set in the secured template area. If an initial value is specified for an element of a dynamic structure variable in the source program, the initial value specified in step S25 is set in an area in the template corresponding to that element. If the initial value is not specified, hexadecimal 8b8b8b8b is set in the same area. This value is used to check whether a value has already been set for an element of a dynamic structure variable in the program while the generated executable program 1 is being executed. Is done.
[0015]
That is, when accessing an element of a dynamic structure variable, if 8b8b8b8b is set, it is determined that a value has not been set for that element (undefined), otherwise a value is set. It can be determined that is set (defined). Although 8b8b8b8b is used here as a value for determining whether or not it is already defined, it is a value that is not used in the program or is used only with a very low probability and has no practical problem. Any value may be used.
[0016]
If the undefined value detection option is not specified in the determination in step S21, the source program 8 is applied to 1/5 or more of the plurality of elements of the dynamic structure variable in step S27. Check whether or not the initial value is specified. If an initial value is specified for an element of 1/5 or more, the process proceeds to step S28, and if it is less than 1/5, the process ends. Note that the value used as the criterion for determination here is not necessarily 1/5, and a value that provides the best execution efficiency of the generated program may be selected.
[0017]
In step S28, a template area used for initializing the dynamic structure variable is secured as in step S22. If a template for a dynamic structure variable of the same type has already been secured, a new template area is not secured. Further, the template area is secured in the static data area of the executable program 1.
[0018]
After securing the template area, in steps S29, S30, and S31, an initial value designated at a location corresponding to an element for which an initial value is designated in the secured template area is set. No value is set for an element whose initial value is not specified. Accordingly, such elements have indefinite values.
7 and 8 show examples of part of the source program. Both are the same except that the beginning of the definition part of the subroutine is shown, and whether or not an initial value is set for the element is partially different.
[0019]
In FIG. 7, the first line is a line indicating the start of subroutine definition. The character string subroutine means the start of the definition, and sub indicates that the name of the subroutine is sub. The var enclosed in parentheses indicates that the subroutine sub has one parameter and its name is var.
All variables defined within a subroutine are dynamic variables. That is, the structure variable defined in the subroutine is a dynamic structure variable. Dynamic variables including dynamic structure variables are dynamically secured in the dynamic data area of the program during execution of the executable program 1. The same applies to the parameters of the subroutine.
[0020]
The second to ninth lines define the dynamic structure variable type. The type on the second line indicates the start of the type definition, and the end type on the ninth line indicates the end of the type definition of the dynamic structure variable. Further, a in the second line indicates that the type name of the type to be defined is a.
The third to eighth lines define the elements constituting the dynamic structure variable. The integer in each line means that the element is an integer type, and the character strings (a1, a2, a3, a4, a5, a6) following the two colons (: :) indicate the element name and follow the element name. The numerical value (1, 2, 3, 4, 5, 6) following the equal sign (=) indicates the initial value for each element.
[0021]
The 10th line indicates that var in the 1st line is a three-element array variable having a structure variable of type a as an element.
The source program shown in FIG. 8 basically has the same definition as that shown in FIG. However, the source program shown in FIG. 8 designates an initial value only for the element a5.
[0022]
When the dynamic analysis unit 9 detects a dynamic structure variable in the syntax analysis unit 9, the source program shown in FIG. 7 determines Y in step S27 in FIG. 5, and the source program shown in FIG. In step S27, N is determined.
When the intermediate language is created by the syntax analysis unit 9, the intermediate language is generated according to the flowchart shown in FIG.
[0023]
First, in step S41, it is checked whether an undefined value detection option is instructed. If an option is instructed, the process proceeds to step S42. If not, the process proceeds to step S43. When the process proceeds to step S43, it is further determined whether or not the element for which the initial value is specified is 1/5 or more of the elements constituting the dynamic structure variable. If the element is 1/5 or more, the process proceeds to step S42. If it is less than 1/5, the process proceeds to step S44.
[0024]
Note that the criterion for determination does not necessarily have to be 1/5 as long as it matches the determination in step S27 of FIG.
In step S42, an intermediate language for initializing the dynamic structure pair variable is generated using the template in which the area is secured in step S22 or S28 of FIG. Here, an intermediate language for transferring the data in the template area collectively to the variable area is generated.
[0025]
FIG. 9 shows an example of the intermediate language generated in step S42. The intermediate language shown in FIG. 9 is generated when the source program shown in FIG. 7 is compiled, or when the source program shown in FIG. 8 is compiled by specifying an undefined value detection option.
The first and third lines mean that the second line is repeatedly executed while changing the variable i from 1 to 3. CMOVE in the second line means that data transfer of a specified number of bytes is performed at a time between two specified areas. The first parameter of CMOVE indicates the data transfer destination, the second parameter indicates the transfer source, and the third parameter indicates the number of bytes of transfer data.
[0026]
In general, a CPU of a computer is provided with an instruction for transferring continuous data at once from one area to another area, and the CMOVE intermediate language is converted into such an instruction by the code generator 11.
FIG. 10 shows an example of the intermediate language generated in step S44. The intermediate language shown in FIG. 10 is generated when the source code shown in FIG. 8 is compiled without specifying the undefined value detection option.
[0027]
The first and fourth lines mean that the second to third lines are repeatedly executed while changing the variable i from 1 to 3. The second line means that the numerical value 5 is loaded into the register r. The third line means that the value stored in the register r (5 in this case) is stored in the structure element a5 of the i-th array element of var, which is an array of dynamic structure variables.
[0028]
In general, the CPU of a computer has a load instruction for setting a value in a register and a store instruction for storing a value in a memory, or an instruction corresponding to these instructions. The intermediate language shown in FIG. The unit 11 converts these instructions.
In general, when comparing the data transfer by the CPU batch data transfer instruction and the data transfer by the load / store instruction, if the amount of data to be transferred is small, the transfer by the load / store instruction is executed faster. However, when transferring a large amount of continuous data, the batch data transfer instruction is executed at a higher speed.
[0029]
With the mechanism of the compiler 6 described above, particularly the mechanism of the syntax analysis unit 9, it is possible to set initialization of dynamic structure variables at high speed and detect undefined values.
If the dynamic data area of the program is realized by calling the library 2 when the program is executed, the contents of the template area are copied to the area of the dynamic structure variable in the library area allocation routine. You may make it do. When the executable program 1 calls the library, information necessary for determining the size of the area to be acquired (the size itself or the size of the structure and the number of array elements of the array including the structure) and the template Pass the address to the library as a parameter. The library that receives these parameters acquires an area and initializes it by repeatedly copying the contents of the template existing at the designated address to the acquired area. This makes it possible to reduce the size of the executable program 1.
[0030]
【The invention's effect】
According to the present invention, the initial value can be set together with the type definition of the dynamic structure variable, and the creator of the source program does not need to write the value every time the structure variable is acquired. .
Since only one template is created for one type, the size of the executable program can be reduced.
[0031]
Further, since the code that is executed by the batch data transfer instruction is generated for copying the contents of the template to the dynamic variable area, the dynamic structure variable can be efficiently initialized.
Also, instead of using a template all the time, code that sets an initial value without using a template when there are only a few elements that specify an initial value among the elements that make up a dynamic structure variable. Therefore, it is possible to create an executable program with high execution efficiency.
[0032]
In addition, the process of copying the contents of the template to the acquired area is performed together with the acquisition of the dynamic area in the dynamic area acquisition routine of the library shared by a plurality of executable programs, thereby reducing the size of the executable program. It can be made smaller.
Furthermore, it is possible to check whether or not an initial value is set for any element of the dynamic structure variable, and it is possible to detect all undefined values.
[Brief description of the drawings]
FIG. 1 is a diagram showing execution of a program on a computer.
FIG. 2 is a diagram showing a program development environment on a computer.
FIG. 3 is a diagram showing the structure and operation of a compiler.
FIG. 4 is a diagram showing the action of a linker.
FIG. 5 is a first diagram showing a part of processing in a structure analysis unit.
FIG. 6 is a second diagram showing a part of processing in the structure analysis unit.
FIG. 7 is a first diagram showing an example of a source program.
FIG. 8 is a second diagram showing an example of a source program.
FIG. 9 is a first diagram illustrating an example of an intermediate language.
FIG. 10 is a second diagram showing an example of an intermediate language.
[Explanation of symbols]
1. . . 1. Executable program . . Library 3. . . OS (Operating System)
4). . . Computer 5. . . Editor 6. . . Compiler 7. . . Linker8. . . Source program 12. . . Object program

Claims (4)

A compiling device that inputs a source program from a storage device and outputs an object program to the storage device ,
An initial value setting area for storing the initial value to be set in the dynamic variable when the ratio of the elements for which the initial value is specified among the elements constituting the dynamic structure variable is greater than or equal to the predetermined value. Initial value setting area acquisition means for statically acquiring in the output object program;
Initial value setting area initializing means for setting an initial value for the acquired initial value setting area;
When the dynamic variable area is dynamically acquired at the time of program execution , among the elements that make up the dynamic structure variable, the ratio of the elements for which the initial value is specified is greater than or equal to the predetermined value. A code for copying the contents of the initial value setting area to the dynamically acquired area is generated in the object program. Otherwise, a value is set only for an element for which an initial value is specified. A compiling device comprising: variable initialization object code generation means for generating code in the object program .   2. The compiling apparatus according to claim 1, wherein said initial value setting area acquisition means acquires only one initial value setting area for one type of dynamic structure variable. 3. The compiling apparatus according to claim 1, further comprising means for setting a predetermined value for detecting an undefined value as an initial value in each element of the dynamic structure variable. On the computer,
  Read the source program from the storage device,
  An initial value setting area for storing the initial value to be set in the dynamic variable when the ratio of the elements for which the initial value is specified among the elements constituting the dynamic structure variable is equal to or greater than the predetermined value. Statically acquired in the object program to be output,
  Set an initial value for the acquired initial value setting area,
  When the dynamic variable area is dynamically acquired at the time of program execution, among the elements that make up the dynamic structure variable, the ratio of the elements for which the initial value is specified is greater than or equal to the predetermined value. A code for copying the contents of the initial value setting area to the dynamically acquired area is generated in the object program. Otherwise, a value is set only for an element for which an initial value is specified. Generate code in the object program,
  Output the generated object program to the storage device
  A computer-readable recording medium on which a program is recorded.

Images