JPH11265290A - Program loading method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To make a program possible to be loaded and executed even in a system not provided with memory resources enough to develop the file of a relative address form on a memory by selecting and loading a storage medium storing object codes in the order of an address at the time of loading. SOLUTION: A file conversion processing 101 receives the file 100 of the relative address form as input and outputs a successively convertible relative address form file 104. That is, records are prepared first, the prepared records are gathered as the file 104 and outputted. A successive arrangement loader 102 reads the successively convertible relative address form file 104 and outputs the program in an absolute address form. The program 108 in the absolute address form can be directly arranged on the memory and executed. In the case that the plural input files 104 are present, a successive loading processing is performed and they are turned the single program 108 of the absolute address form and outputted.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] 1. Field of the Invention [0002] The present invention relates to a method of loading a program in using a computer.

[0002]

2. Description of the Related Art An object code format of a program includes an absolute address format and a relative address format. The absolute address format is a format in which a memory reference or a branch destination is specified by an absolute address. Only programs in absolute address format can be directly executed by the CPU.

[0003] The relative address format is a format in which a memory reference or a branch destination is specified by a relative method such as a symbol. Relative address format programs cannot be executed directly by the CPU. Convert to absolute address format through loader and execute. Unlike the absolute address format, the relative address format program can specify the location address when loading. for that reason,
The advantage is that you can freely select the location on the memory,
Has been adopted in many computer systems.

[0004] A program file in the relative address format includes a program body and relocation information. The relocation information is information referred to by the loader when converting the program body into the absolute address format. An overview is given in "System Program I, Chapter 5 Loader" by John J. Donovan, published by the Japan Computer Association. COFF is the most widely used program file format
(Common Object File Format). See AT & T UNIX SYSTEM V Programmers R for details.
eference Manual Chapter 4 (File Formats) ".

[0005]

A loader for arranging a relative address format file reads all the files, expands them on a memory, and converts them into an absolute address format file. Since the relative address format file has the relocation information, the size is often several times the size of the absolute address format file. Therefore, in order for the loader to operate, it is necessary to secure a working memory several times as large as the program body. If the working memory cannot be secured, the program file cannot be loaded and the program cannot be executed.

An object of the present invention is to enable a program to be loaded and executed even in a system which does not have a memory resource capable of expanding a file of a relative address format in a memory.

[0007]

Means for Solving the Problems (1) A conventional relative address format file is converted into a program format including a set of records including an instruction body and instruction relocation information as elements.

(2) The loader takes out one record at a time from the file in the program format of (1), converts it into the absolute address format, and arranges it on the memory.

(3) If the file of the program format of (1) cannot be stored in a single medium, the file is divided and stored in a plurality of media.

(4) The loader sequentially loads the files divided as described in (3) above, reconfigures them into a single program, and arranges them on the memory.

[0011]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 1 shows the overall configuration. The file conversion processing 101 receives a file 100 in a relative address format as an input and outputs a sequentially convertible relative address format file 104. The sequentially convertible relative address format file 104 can be stored in a disk device or an IC card. If the data cannot be stored in a single medium, the data can be divided and stored in a plurality of media.

The most common format of the relative address format file 100 is the COFF format. The successively convertible relative address format file 104 is a file having a format including a set of records including an instruction body and instruction relocation information as elements. The file conversion process 101 first creates a record in 102. There are two methods for creating a record: a method using relocation information as a unit and a method using an instruction as a unit.
Next, the created records are output together to a file.
If the file does not fit on a single medium, the file is divided into a plurality of files by file division processing 103. When an IC card is used as a medium for storing a file, the storage capacity is strictly limited, so that it is often necessary to divide the file.

The sequential placement loader 102 reads the sequentially convertible relative address format file 104 and outputs an absolute address format program. The program in the absolute address format can be directly arranged on the memory and executed. You can write it directly to memory or save it to a file. If there are a plurality of input files, they are sequentially loaded and output as a single absolute address format program. The sequential placement loader 104 first reads the file header and determines the instruction placement address. Based on this address, record arrangement processing 107 is performed to generate an absolute address format program 108.

FIG. 2 shows an example of an object file in a relative address format. This example is simplified compared to a common COFF format file for explanation.
The relative address format file 200 is roughly divided into a file header 201, a code area 202, and a relocation information area 20.
It consists of three parts, seven. The file header 201 describes information such as file attributes and area arrangement. In the code area 202, instruction strings 203, 204, 205, and 206 are written. Some instructions are not subject to relocation processing. 204 and 205 are instructions not to be relocated. The instruction written in the code area becomes an absolute address instruction as it is. 203 and 206 are instructions to be relocated.
These are converted into absolute address instructions at the time of loading based on the relocation information 208 and 209.

There are two types of relative format file formats that can be sequentially converted. One is to create a record in units of instructions, and the other is to create a record in units of relocation information. A method for creating a record in units of relocation information will be described.

FIG. 3 shows an example of a sequentially convertible relative format file 300 for creating a record in units of relocation information. The file 300 includes a file header 301 and a record list 302. The file header 301 describes the file identification information and the offset at which the record is arranged. The record list 302 is a list in which many records such as 303 and 304 are arranged in order.

The record includes a relocation target instruction corresponding to the relocation information and a non-relocation target instruction. Although there is always one relocation information and one relocation target instruction, there may be no relocation non-relocation target instruction, or there may be more than one. In the case of the record 303, there are relocation information 208 and a relocation target instruction 203, and there are two non-relocation target instructions 204 and 205. Record 304 is
It has the relocation information 209 and the relocation target instruction 206, but has no relocation target instruction.

Since each record holds the relocation information and the relocation target instruction in the record, the loader can arrange the instruction as soon as one record is read.

Reference numerals 400 and 403 in FIG. 4 show an example of division of a sequentially convertible relative format file for creating a record in units of relocation information. The sequentially convertible relative format file can be divided into arbitrary files in units of records. The record list 302 is divided into a record list 402 and a record list 405, which are elements of a file 400 and a file 403, respectively.

The file header shows the file identification information and the offset of the code area. The file header 401 describes only file identification information. Since the file 400 is the first file, instructions may be arranged from the beginning of the code area. File header 404
Describes the file identification information and the code offset. Since the file 403 is the second file, it must be arranged with a space for the instruction of the first file. It can be seen from the header 404 that the instruction should be arranged from the point of 16 bytes from the head of the code area.

FIG. 5 shows a file conversion processing procedure for generating a sequentially convertible relative format file for creating a record in units of relocation information. First, a file in a relative address format is read at 500 and expanded in a memory. Next, at 501, the relocation information area is checked, and the number of registered relocation information entries is checked. Create a record in the relative format that can be sequentially converted by the number of the entries. In 502, the entry in the relocation information area is scanned, and the relocation information is written into a record in a sequentially convertible relative format. Next, the instruction in the code area is written in the record. At this time, the instruction to be relocated and the instruction not to be relocated are handled differently. The instruction to be relocated is written 503 in each record one instruction at a time. The instruction not to be relocated is incorporated in 504 in the same record as the immediately preceding instruction to be relocated. Therefore, a plurality of instructions not to be relocated may be written, or they may not be written at all.
Eventually all instructions are written to the record.

At 505, the size of the sequentially convertible relative format file is estimated. Since the size of the header is fixed, the size of the entire file can be estimated from the memory occupied by the record. If the file size is within the storage size of the medium to be written, it is treated as a single file as it is. If the record list does not fit in the medium, the record list is divided at an appropriate place in 506 and divided into a plurality of files. 507 calculates the instruction offset value of each file,
Create a header. Finally, at 508, the record is written to the file.

In the case of a confidential program, the process 50
7 describes the permission conditions in the header. Further processing 508
Encrypts the record with and writes it to the file. Encryption may be performed on a record basis or on the entire record list.

FIG. 6 shows a processing procedure of a sequential placement loader for arranging a sequentially convertible relative format file for creating a record in units of relocation information. At 600, read the file header in the sequential layout relative format. At 601, the file attributes in the file are referenced. If the header has permission information, the condition is checked, and if the condition is met, the processing is continued. Then check if the file is split. If it is a single file, the process is continued. If the file is divided into multiple files, check the instruction offset.
In the subsequent processing 107, instructions are arranged from a position shifted by an offset in the code area.

In step 602, records are sequentially scanned. If the record is encrypted, it is decrypted here. When all records have been scanned, the process is completed. First, at 603, record relocation information is extracted. Next, in step 604, the instruction to be relocated is extracted, and an absolute address instruction is generated together with the relocation information read earlier. The instruction generated in 605 is arranged in the code area of the memory. Next, an instruction not to be relocated is checked at 606. If there is no instruction to be relocated, the process moves to the next record. 607,608 if there are non-relocation instructions
And place it in the code area of the memory as it is. After writing the command, the process moves to the next record.

When the processing is completed, the arrangement information of the memory code area is checked. In the case of a single file, all the code areas should have already been placed. Notify the system that the program is ready to run. If the file is divided into a plurality of files, not all of the code areas are already arranged. When it is confirmed that all the instructions have been arranged, the system is notified that the program can be executed.

Reference numeral 700 in FIG. 7 shows an example of a sequentially convertible relative format file for creating a record in instruction units.
File 700 has file header 701 and record list
Consists of 702. The file header 701 describes the file identification information and the offset at which the record is arranged. The record list 702 is a list in which many records such as 703, 704, 705, and 706 are arranged in order.

A record is composed of relocation information and an instruction.
When the instruction is the instruction to be relocated, the instruction has the relocation information and the instruction. However, the instruction that is not the instruction to be relocated has empty relocation. The records 703 and 706 have relocation information 208 and 209, respectively, because the instructions 203 and 206 are relocation target instructions. In the records 704 and 705, the relocation information is empty because the instructions 204 and 205 are non-relocation instructions.

Since each record holds information necessary for arrangement in the record, the loader can arrange an instruction as soon as one record is read.

Reference numerals 800 and 803 in FIG. 8 show examples in which a sequentially convertible relative format file for creating a record for each instruction is divided. The sequentially convertible relative format file can be divided into arbitrary files in units of records. The record list 702 is divided into a record list 802 and a record list 805, which are elements of a file 801 and a file 803, respectively.

The file header shows the file identification information and the offset of the code area. The file header 801 describes only file identification information. Since the file 800 is the first file, the instructions may be arranged from the beginning of the code area. File header 804
Describes the file identification information and the code offset. Since the file 803 is the second file, it must be arranged with a space for the instruction of the first file. It can be seen from the header 804 that the instruction should be arranged from the point of 16 bytes from the head of the code area.

FIG. 9 shows a processing procedure of file conversion for generating a sequentially convertible relative format file for creating a record for each instruction. First, a file in a relative address format is read at 900 and expanded in a memory. Then, 90
Check the code area with 1, check the number of registered instructions,
Create as many records as there are entries. Register the command in the record generated in step 902. In 903, the entry in the relocation information area is scanned to check whether there is relocation information corresponding to the instruction. If there is no information, the instruction is a non-relocation target instruction. Leave the relocation information column empty. If there is information, it is a relocation target instruction. In step 904, the relocation information is registered in the record.

When all the instructions are finally written in the record, the size of the sequentially convertible relative format file is estimated at 905. Since the size of the header is fixed, the size of the entire file can be estimated from the memory occupied by the record. If the file size is within the storage size of the medium to be written, it is treated as a single file as it is.
If the record list does not fit in the medium, the record list is divided at an appropriate place in 906 and divided into a plurality of files. 907
Calculates the instruction offset value for each file and creates a header. Finally, at 908, the record is written to the file.

In the case of a confidential program, the process 90
7 describes the permission conditions in the header. Further processing 908
Encrypts the record with and writes it to the file. Encryption may be performed on a record basis or on the entire record list.

FIG. 10 shows a processing procedure of a sequential placement loader for arranging a sequentially convertible relative format file for creating a record for each instruction. Read file header at 1000. Refer to the file attribute reference in 1001. If the header has permission information, the condition is checked, and if the condition is met, the processing is continued. Then check if the file is split. If it is a single file, the process is continued. If the file is divided into multiple files, check the instruction offset. In the subsequent processing 107, instructions are arranged from a position shifted by an offset in the code area.

Next, in step 1002, records are sequentially scanned. If the record is encrypted, it is decrypted here. When all records have been scanned, the process is completed. First 1003
Check the record relocation information to see if the information is written. If the information is empty, the instruction is directly written into the code area of the memory at 1004. If there is relocation information, the relocation information is read at 1005, and the relocation target instruction is read at 1006. An absolute address instruction is generated by combining the relocation information and the instruction. The generated instruction is written in the code area of the memory at 1007.

When all records have been processed, the arrangement information of the code area in the memory is checked. In the case of a single file, all the code areas should have already been placed. Notify the system that the program is ready to run. If the file is divided into a plurality of files, not all of the code areas are already arranged. When all the files are loaded and it is confirmed that all the instructions are arranged, the system is notified that the program is executable.

FIG. 11 shows an execution memory image with relocation information. All programs are converted to an executable absolute address format and placed in memory. one time,
When the data is converted into the absolute address format, the relocation information is lost, so that the relocation can no longer be performed. To relocate code in absolute address format, the relocation information must be stored separately.

Reference numeral 1101 denotes a memory image on the system. 1103 is a code area of an executable application. 1104, 1105, 1106, 1107 are instructions in the code area. Each instruction is a placed executable instruction. A memory area 1109 stores instruction relocation information. 208 and 209 are relocation information. An area 1110 stores additional information other than the relocation information.
1109 and 1110 are not required for running the program. Used to relocate a program in a code area and load it on another system.

Reference numeral 1102 denotes a system area, and 1111 denotes an area for another application.

FIG. 12 shows a procedure for loading from an execution memory image with relocation information to another terminal.
At 1201, the code area of the memory of the original terminal is examined. If there is no command, the process ends. At 1202, an instruction is taken out of the code area, and at 1203, it is checked whether or not there is corresponding relocation information. If there is no relocation information, in step 1205, the instruction is transferred to the load destination terminal. If there is the rearrangement information, the instruction is rearranged in step 1204, and when the processing is completed, the instruction is transferred to the load destination terminal.

When the processing of the instruction is completed, the process proceeds to 1207. Consider whether you intend to load the program loaded on the new terminal again on another terminal. If there is no intention to load the data to another terminal, the processing may be ended as it is. If you intend to load it, you cannot load it to another terminal without relocation information. At 1007, the relocation information is transferred to the terminal to be loaded.

FIG. 13 shows an example of a relocated program file. The file is composed of a file header 1302 describing attribute information of the program, and a plurality of program bodies 1305. The program main body 1305 has a load address field 1304 and a rearranged object field 1306. A relocated object is one in which an object is relocated by designating a specific address, and can be directly written to a memory and executed. The load address 1304 is a field describing the address.

The procedure for loading the relocated program is simple. First, the load address of the program is obtained from the system. Next, the load address field in the file is searched for a match with the address obtained from the system. If there is no match, you cannot load. If there is a match, the relocated object may be extracted and written to the load destination memory as it is. Requires little working memory.

Although the load address is limited to the one prepared in advance, the computer system generally manages the memory by dividing the memory into blocks of a certain size. Should be prepared. In the case of a device such as a flash memory, the address to be prepared is further limited because the specification of the chip and the delimitation of the block are determined.

[0046]

According to the present invention, a program can be loaded and executed even in a system having no memory resources capable of expanding a file of a relative address format on a memory.

[Brief description of the drawings]

FIG. 1 is an overall configuration of a program loading method.

FIG. 2 is an example of a relative address format file.

FIG. 3 is an example of a sequentially convertible relative format file in which records are created in units of relocation information.

FIG. 4 is an example of division of a sequentially convertible relative format file in which records are created in units of relocation information.

FIG. 5 shows a file conversion processing procedure for generating a sequentially convertible relative format file for creating a record in units of relocation information.

FIG. 6 is a processing procedure of a loader that arranges a sequentially convertible relative format file that creates a record in units of relocation information.

FIG. 7 is an example of a sequentially convertible relative format file for creating a record in instruction units.

FIG. 8 is an example of division of a sequentially convertible relative format file that creates a record for each instruction.

FIG. 9 is a file conversion processing procedure for generating a sequentially convertible relative format file for creating a record for each instruction.

FIG. 10 shows a processing procedure of a loader for arranging a sequentially convertible relative format file for creating a record in instruction units.

FIG. 11 is a diagram showing an execution memory image with relocation information on the system.

FIG. 12 is a processing procedure for loading a program from an execution memory image with relocation information to another system.

FIG. 13 shows an example of a relocated program file.

Claims (2)

[Claims] In a program loading method, an object code linked in advance by designating a load address is divided and stored in a plurality of storage media. A program loading method comprising selecting a stored storage medium and loading a program. 2. The program loading method according to claim 1, wherein a program file is created using a flash block boundary or a page boundary of a target machine as a link address.