JPH10283194A - Program linkage editor method and machine readable record medium recording program linkage editor program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To reduce the number of instructions which are needed for access by having an inherent global pointer in each task. SOLUTION: An assembler program is inputted (S101), a list of global pointers which are used in a global pointer section is generated (S102), an offset to a label is decided for each section of the global pointer section (S103), the size of each section is decided in each section (S104), plural sections that are related to each global pointer are continuously arranged and a leading address of the section is decided (S105), the value of each global pointer is decided (S106), an address and an offset based on a relocation address are generated and the relocation address is calculated (S107), and a generated executable object is outputted (S108).

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a program combination editing method and a machine-readable recording medium recording a program combination editing program. Technology that makes it possible to have

[0002]

2. Description of the Related Art When a part of the logical address space can be accessed with a small number of instructions, the total number of instructions can be reduced by each multitask task having a unique global pointer. A microcomputer architecture that can be used is conventionally known. An example of a microcomputer having such characteristics is MI
There is a PS architecture (Reference: Erin Farquhar
and Philip Bunce "The MIPS Programmer's Handbook",
1994).

A specific instruction processing operation in the MIPS architecture will be described with reference to FIG.
The instruction sequence enclosed by the upper right square in FIG. 17 stores the value of the variable on the memory designated by the 32-bit address in the register t.
This is an instruction sequence for loading to 0. The lui AT, 0x1234, which is the first instruction in this instruction sequence, is a general register (GR) labeled with the immediate value "0x1234"("0x" indicates hexadecimal notation) and "AT". Is an instruction to be stored in the upper 16 bits. The instruction in the second column, lw t0,0 (AT), is a value stored in a memory address corresponding to a value obtained by adding an offset “0” to a value stored in a general register with a label “AT”. Is stored in the general register labeled “t0”. Therefore, the memory address “0x1
The value stored in “2340000” is stored in the general register labeled “t0”.

The source program of the assembler is shown in FIG.
Although it can be described by one instruction apparently like an instruction sequence enclosed by a square at the upper left of FIG. 7, two instructions at the upper right of the figure are actually generated as machine language instructions.

FIG. 18 is a diagram for explaining an instruction processing operation when access is possible using a global pointer. The instruction sequence enclosed by the upper right square in FIG. 18 is an instruction sequence for loading the value of the variable on the memory specified by the 32-bit address into the register t0.
Lui gp, 0x1234, which is the first instruction in this instruction sequence, is a general register (GR) labeled with the immediate value “0x1234” (“0x” indicates hexadecimal notation) and “gp”. Is an instruction to be stored in the upper 16 bits. This instruction is generally defined at the beginning of the program. The instruction described below, lw t0,0 (gp), is stored in a memory address corresponding to a value obtained by adding an offset “0” to the value stored in the general register of the label “gp”. The stored value is stored in the general register labeled “t0”. Therefore, the memory address “0x1
The value stored in “2340000” is stored in the general register labeled “t0”.

In the MIPS architecture, the range of the relative address of 16 bits from the base address of the register can be accessed by one instruction by defining the global pointer at the beginning of the program as described above. The total number of instructions can be reduced.

Conventionally, there has been a combination editing (linkage) method for a microcomputer having the above-described architectural features, which corresponds to one global pointer in one combination editing unit. FIG. 19 is a diagram showing a flowchart of a conventional combination editing method. First, a relocatable file is input (step S301). Next, for each section,
The offset is determined (step S302), and the size of each section is determined for each section (step S3).
03). Subsequently, a plurality of sections related to the global pointer are successively arranged, and the start address of the section is determined (step S304). Subsequently, the value of the global pointer is determined (step S305), an address and an offset are generated based on the relocation information, and the relocation address is calculated (step S306). Finally, the generated matter-possible object is output (step S
307).

Next, this conventional method will be described in detail with specific examples. FIG. 20 shows an assembler program shown to explain a specific example of the conventional method.

In FIG. 1A and FIG. s
The bss pseudo-instruction declares a variable without an initial value that is referenced using a global pointer. For example, FIG.
".Sbss a, 4" shown in (1) is an instruction to allocate a 4-byte area to the label a. on the other hand,. The sdata pseudo-instruction declares a variable with an initial value that is referenced using a global pointer. For example,
".Sdata x .word" shown in FIG.
"0" is an instruction for allocating an area of one word (4 bytes) to the label x and setting its initial value to 0.

In addition,. The text section is a macro for calculating an offset from a global pointer as an operand of an instruction, and% gpoffs (lab
el) is provided. This is because the label for which the combination editing program determines the value at the time of the final combination editing and the. This is a macro for calculating a relative offset with respect to a variable allocated in a section having a name allocated by the gpsection pseudo instruction.

A description will be given of a conventional method of combining and editing two programs shown in FIGS. First, in step S301, the relocatable file shown in FIG. 20 is input. Here, relocatable (relocat
"able" means a format in which the address designation part is expressed in a form to which modification information that enables address calculation is added at the time of the subsequent link editing processing.

Subsequently, in step S302, an offset is determined for a label in the section for each section described in the input file. FIG. 20 (a)
In the assembler program shown in (1), "x" is described as a label. Further, in the assembler program shown in FIG. 2B, “y” is described as a label. First, for this "x", the section is "s
data ”and the offset is“ 0 ”. FIG.
(A) shows this for easy understanding. Similarly, for “y”, the section is “sda
ta ”, and the offset is“ 0 ”. FIG. 21 (c)
This is shown here to make it easier to understand.

Subsequently, in step S303, the size of each section is determined for each section. FIG.
FIG. FIG. 21D shows the size of the sdata section, and FIG.
In (b). The size of the sdata section is determined. As described above, since the variables are defined by one word in both sections, it is understood that the size is 4 bytes. Further, FIG.
In (a). FIG. 21F shows the size of the label “a” in the sbss section, and FIG.
At. This is determined for the size of the label “a” in the sbss section. Thus, it can be seen that both sections are 4 bytes in size.

Subsequently, in step S304, a plurality of sections relating to each global pointer are successively arranged, the start address of the section is determined, and in step S3
At 05, the value of each global pointer is determined, and at step S306, an address and an offset based on the relocation information are generated, and the relocation address is calculated.

[0015] After being combined and edited in this manner, it is arranged on the memory as shown in FIG. FIG. 22 shows FIG.
It is a figure for explaining the case where 0 (a) and (b) are combined and edited and arranged in the memory. here,. sbss section and. In the sdata section, variables without initial values and variables with initial values are respectively arranged. Moreover,. sb
ss section and. The sdata section is arranged continuously. When the address allocation of these sections is completed, the link editing program executes the symbol_gp
, An address at a position of 32 KB from the head of the sbss section is allocated. After that, the combined editing program: & in the instruction operand in the text section
The offset corresponding to the gpoffs macro is calculated, and a final machine language instruction is generated.

As described above, in the conventional combined editing method, the section to which the data to be referred to belongs is determined by using the global pointer, and the size of the section and the finally arranged address are determined during the combined editing. After that, the initial value of the symbol which is the promise of storing the initial value of the global pointer is determined, and then the offset value of the memory access instruction referring to the global pointer is determined.

[0017]

However, in the conventional combined editing method, only one data section referred to by using the global pointer can be provided in the combined editing unit.

When an application program is created using a multitasking operating system, a unique global pointer is provided for each task, and when switching tasks, the value of the global pointer is changed to a value corresponding to the task. In some cases, the number of instructions required for access can be reduced by making the change. In particular, the method of sharing one global pointer for all tasks with such a data capacity is effective when the data exceeds the maximum capacity limited by the architecture. In the conventional combined editing method, it is difficult to create a program that has a plurality of global pointers in a combined edit unit.

The present invention has been made in view of the above circumstances, and an object of the present invention is to make it possible to have a unique global pointer for each task in the creation of a program for linking and editing all at once. Accordingly, it is an object of the present invention to provide a program combination editing method and a machine readable recording medium recording a program combination editing program which can reduce the number of instructions required for access.

[0020]

According to a first aspect of the present invention, there is provided a method for connecting and editing a plurality of assembler source programs including a global pointer section. For the global pointer section, data without an initial value and data with an initial value are individually arranged for each global section, and a unique global pointer is provided for each global pointer section included in the plurality of assembler source programs. Outputting the executable object.

In general, when the unit of combined editing is large and only one global pointer section can be used as a whole, it may not be possible to store all variables in the global pointer section. According to the configuration of the present invention, when the independence between sub-modules in a program is high, as represented by a multi-task program, a unique global pointer is provided for each task and used for each task. Frequent variables can be arranged so that they can be accessed by their global pointers, and the number of instructions as a whole can be reduced.

According to a second aspect of the present invention, there is provided a method of connecting and editing a plurality of assembler source programs including a global pointer section, wherein the plurality of global programs specified in the assembler source program are combined. A relative offset from the value of the global pointer relating to the pointer section is obtained, and an address in each of the global pointer sections is calculated using the relative offset value. And outputting an executable object provided with a global pointer unique to each global pointer section included in the plurality of assembler source programs.

According to the configuration of the present invention, when the independence between sub-modules in a program is high, as represented by a multitask program, each task has a unique global pointer, and each task has a unique global pointer. Variables that are frequently used can be arranged so that they can be accessed by the respective global pointers, so that the total number of instructions can be reduced.

According to a third aspect of the present invention, there is provided a combined editing method of a program for performing combined editing on an assembler source program including a plurality of global pointer sections, wherein the plurality of global pointers specified in the assembler source program are combined. A relative offset from the value of the global pointer relating to the pointer section is obtained, and an address in each of the global pointer sections is calculated using the relative offset value. And outputting an executable object provided with a global pointer unique to each global pointer section included in the plurality of assembler source programs.

Even if the data handled by one task is large, if the frequency of reference to the variable differs for each part, the number of instructions can be reduced by replacing the global pointer for each part and accessing. is there.
According to the configuration of the present invention, such a program can be combined and edited.

According to a fourth aspect of the present invention, there is provided a method for connecting and editing a plurality of assembler source programs including a global pointer section, wherein the plurality of assembler source programs include a global pointer section. Find relative offsets from the values of the global pointers related to the pointer section,
An address in each of the global pointer sections is calculated using the relative offset value, and data without an initial value and data with an initial value are individually arranged for each global pointer section, and are included in the plurality of assembler source programs. And outputting an executable object provided with a unique global pointer for each global pointer section.

Even if the task is divided into several parts, even if the size of the data to be handled does not fit in one global pointer section, the number of instructions may be reduced by allocating a plurality of registers as global pointers. . According to the configuration of the present invention, such a program can be combined and edited.

According to a fifth aspect of the present invention, there is provided a method for connecting and editing a plurality of assembler source programs including a global pointer section, the method comprising: Calculating an offset from a value of a global pointer relating to the plurality of global pointer sections specified in the assembler source program; data without an initial value and data with an initial value are individually arranged for each global pointer section; An executable object in which a unique global pointer is provided for each global pointer section included in the assembler source program is output.

[0029] In the configuration of the fourth aspect of the present invention,% gpo
A global pointer corresponding to a label must be specified as an argument to the ffs macro.
According to the configuration described above, there is an effect that it is not necessary.

A sixth aspect of the present invention is a method for connecting and editing a plurality of assembler source programs including a global pointer section, wherein the assembler program is input and the global pointer section is input to the global pointer section. A list of global pointers used in each section of the global pointer section, determine an offset for a label in the section for each section, determine a size of each section for each section, Multiple sections related to pointers are arranged consecutively, the start address of the section is determined, the value of each global pointer is determined, the address and offset based on the relocation information are generated, and the relocation address is calculated. , Output of generated executable object And performing.

According to the configuration of the present invention, a unique global pointer is provided for each task. Therefore, a variable frequently used for each task can be arranged so as to be accessed by each global pointer. This makes it possible to achieve a reduction in the number of instructions as a whole.

In order to achieve the above object, the invention of claim 7 is a machine readable recording medium which records a combined editing program of a program for performing combined editing on a plurality of assembler source programs including a global pointer section. In the global pointer section, data without an initial value and data with an initial value are individually arranged for each global section, and a global pointer unique to each global pointer section included in the plurality of assembler source programs. Output the executable object provided with.

According to another aspect of the present invention, there is provided a machine-readable recording medium recording a combined editing program of a program for performing combined editing on a plurality of assembler source programs including a global pointer section,
Inputting the assembler program, creating a list of global pointers used in the global pointer section, and determining, for each section of the global pointer section, an offset with respect to a label in the section When,
Deciding the size of each section, arranging a plurality of sections related to each global pointer in succession, determining the start address of the section, and deciding the value of each global pointer for each section And a step of generating an address and an offset based on the relocation information to calculate a relocation address, and a step of outputting the generated executable object.

[0034]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Embodiments of a program combination editing method and a machine readable recording medium recording a program combination editing program according to the present invention will be described below in detail with reference to the drawings.

First Embodiment A first embodiment of the present invention will be described with reference to the drawings. FIG. 1 is a flowchart illustrating a program combination editing method according to the present embodiment. In the program editing method according to the present embodiment, first, a relocatable file is input (step S101). Subsequently, a list of global pointers to be used is created (step S102). Subsequently, an offset is determined for a label in the section for each section (step S103), and a size of each section is determined for each section (step S104).
Subsequently, a plurality of sections related to each global pointer are arranged consecutively, and the head address of the section is determined (step S105). Subsequently, the value of each global pointer is determined (step S106), an address and an offset are generated based on the relocation information, and the relocation address is calculated (step S107). Finally, the generated executable object is output (step S10).
8).

Next, this embodiment will be described in detail with reference to specific examples. First, a relocatable file is input in step S101. FIG. 2 shows an assembler program used for explaining a specific example of the present embodiment. In FIG. s
bss pseudo-instruction and. The sdata pseudo-instruction is shown in FIG.
This is the same as the description for 0. The difference from FIG. 20 is that
.. At the beginning of the assembler program in FIG. The gpsection pseudo-instruction is added. To specify the data section to be accessed via the global pointer, as an assembler pseudo-instruction, The gpsection pseudo instruction is provided. Only once for one assembler source. A section name can be assigned by the gpsection pseudo-instruction. FIGS. 7A and 7B are stored in individual files.

Subsequently, in step S102, a list of global pointers to be used is created. In the global pointer used in the present embodiment, the global pointer of the global pointer section “task1” shown in FIG. 2A is set to “_gp_task1”.
The global pointer section “ta” shown in FIG.
The global pointer of “sk2” is set to “_gp_task”
2 ".

Subsequently, in step S103, an offset is determined for a label in the section for each section described in the input file. In the assembler program shown in FIG. 2A, “x” is described as a label. Further, in the assembler program shown in FIG. 2B, “y” is described as a label. First, for “x”, “_data” indicating that the section is a data section is added to the global pointer section “task1” defined above, the section name is “task1_data”, and the offset is “0”. I do. FIG. 3A shows this for easy understanding. Similarly, for “y”, “_data” indicating that the section is a data section is added to the global pointer section “task2” that defines the section above, and the section name “tas”
k2_data ”and the offset is“ 0 ”. This is shown in FIG. 3B for easy understanding.

Subsequently, in step S104, the size of each section is determined for each section. First, FIG.
In the assembler program shown in (a), "a" is described as a label. In the assembler program shown in FIG. 3B, “b” is described as a label. First, regarding this “a”, the global pointer section “tas” defining the section above
k1 ”. "_bs" indicating that the section is a bss section
s ”to add the section name“ task1_bss ”
And the area size is 4. This is shown in FIG. 3C for easy understanding. Similarly, for "b", the section is added to the global pointer section "task2" defined above. “_bss” indicating that the section name is a section name “t”
ask2_bss ”, and the area size is 4. FIG.
This is shown in (d) for easier understanding.

Therefore, the global pointer section t
When the size is determined for each section related to ask1, for "task1_bss", the attribute is "_gp_task1" as a global pointer, and
Size is 4. For “task1_data”, the attribute “_gp_t” is used as a global pointer.
ask1 ”and the size is 4. further,. t
The size of the ext area is 8. FIG. 3E shows this for easy understanding. On the other hand, task
Determine the size for each section about 2,
“Task2_bss” has an attribute “_gp_task2” as a global pointer and a size of 4. For “task2_data”, the attribute “_gp_task” is used as a global pointer.
2 "and the size is 4. further,. text
The size of the area is 8. FIG. 3 (f) shows this for easy understanding.

FIG. 4 shows the size of the result for each section. In this chart,. Since the text section is present in both FIG. 3 (e) and (f),
The total size is 16.

Subsequently, in step S105, a plurality of sections relating to each global pointer are successively arranged, and the head address of the section is determined.
At 06, the value of each global pointer is determined. Accordingly, the initial value of the global pointer “_gp_task1” is 0x00008000, and the initial value of the global pointer “_gp_task1” is 0x000.
08008.

Subsequently, in step S107, an address and an offset based on the relocation information are generated, and the relocation address is calculated. The calculation of the relocation address is shown in FIG.
As shown in (2), it is calculated by (label name)-(initial value of corresponding global pointer). In this step, after determining the addresses of all the labels and the initial values of the global pointers, an offset value is obtained in accordance with the calculation formula shown in FIG. 6, and the offset value is directly embedded in the output executable object.

FIG. 7 shows the memory image after the combined editing. For each assembler source, the initial value of each global pointer is _gp_task1 and _gp
_Task2. For task1_bss and task1_data, continuous areas are allocated, and gp_task is located at a position 32 KB from the top of them.
1 is assigned. Similarly, task2_bss and t
ask2_data is allocated a continuous area,
Gp_task2 is allocated to the position of 32 KB from the head. As described above, the data without the initial value and the data with the initial value are individually arranged for each global section in the global pointer section, so that the global pointer unique to each global pointer section included in the assembler source program is allocated. The pointer is provided.

The instruction code is. It is organized into text sections. When executing these codes,
It is necessary to set the value of the global pointer corresponding to each task in the gp register.

In this embodiment, the description of the initialization of the gp register and the replacement of the gp register accompanying the task switching will be omitted. However, in a normal multitask operating system function, the gp register is used as a task-specific resource. Can be easily realized by assigning

As described above, if the unit of combined editing is large and only one global pointer section can be used as a whole, all variables cannot be stored in the global pointer section. According to the present invention, when the independence between sub-modules in a program is high, as represented by a multi-task program, each task has a unique global pointer. Variables that can be accessed by their global pointers,
A reduction in the number of instructions as a whole can be achieved.

Second Embodiment Next, a program combination editing method according to a second embodiment will be described with reference to FIG. In the present embodiment, the first
For the embodiment of. A plurality of gpsection pseudo-instructions are described in one assembler source program. . sbss,. For the sdata pseudo-instruction and the% gpoffs macro, the. The definition of gpsection becomes valid. FIG.
When the processing shown in the flowchart of FIG. 1 is performed on the assembler source program shown in FIG. 7, the memory image after the combined editing becomes the same as that of FIG. 7 which is the result of the first embodiment. According to the program combination editing method of the present embodiment, even when the data handled by one task is large, if the frequency of reference to the variable is different for each part, the global pointer is replaced for each part and accessed. In some cases, the number of instructions can be reduced. Such programs can be combined and edited.

Third Embodiment Next, a program combination editing method according to a third embodiment will be described with reference to the drawings. This embodiment is different from the first embodiment in that. gpsection pseudo-instruction and% g
The function of the poffs macro is extended to enable a certain program thread to use a plurality of registers as global pointers. Also in the present embodiment, processing is performed as shown in the flowchart of FIG. This embodiment will be described in detail with reference to specific examples. First, a relocatable file is input in step S101. FIG. 9 shows an assembler program used for explaining a specific example of the present embodiment. In FIG. The sbss pseudo-instruction and. The sdata pseudo-instruction is the same as that described with reference to FIG. 20. However, in FIG. gps
There is an action pseudo-instruction. The difference from FIG. 20 is that the assembler program shown in FIG. 9A and FIG. The gpsection pseudo-instruction is added. To specify the data section to be accessed via the global pointer, as an assembler pseudo-instruction, The gpsection pseudo instruction is provided. For one assembler source,
This one time only. A section name can be assigned by the gpsection pseudo-instruction. In addition, FIG.
of. It is assumed that 100 instructions are described in the text area.

Subsequently, in step S102, a list of global pointers to be used is created. In the global pointer used in the present embodiment, the global pointer of the global pointer section “gp1” shown in FIGS. 9A and 9B is “gp1”, and the global pointer section “gp1” shown in FIG. The global pointer of “gp2” is “gp2”.

Subsequently, in step S103, an offset is determined for a label in the section for each section described in the input file. In the assembler program shown in FIG. 9A, “x1”, “x
2, "y1" and "y2" are described as labels. In the assembler program shown in FIG. 2B, “z1” and “z2” are described as labels. First, for “x1”, “_data” indicating that the section is a data section is added to the global pointer section “gp1” that defines the section above,
The section name is “gp1_data” and the offset is “0”. Similarly, for “x2”, the global pointer section “gp” whose section is defined above
"_Data" indicating that this is a data section in "1"
To make the section name “gp1_data”,
The offset is “64”. As for “y1”, “_da” indicating that the section is a data section is added to the global pointer section “gp2” defined above.
ta ”and the section name“ gp2_data ”
And the offset is “0”. As for “y2”, “_data” indicating that the section is a data section is added to the global pointer section “gp2” that defines the section above, and the section name is “gp2_”.
data ”and the offset is“ 64 ”. FIG.
(A) shows this for easy understanding.

Also, for "z1", the global pointer section "gp1" whose section is defined above
Is added to “_data” indicating that the section is a data section, the section name is “gp1_data”, and the offset is “0”. As for “z2”, “_da” indicating that the section is a data section is added to the global pointer section “gp2” defined above.
ta ”and the section name“ gp2_data ”
And the offset is “32”. This is shown in FIG. 10B for easy understanding.

Subsequently, in step S104, the size of each section is determined for each section. First, FIG.
In the assembler program shown in (a), "a",
“B” is described as a label. In the assembler program shown in FIG. 3B, “c” is described as a label. First, for this “a”, the global pointer “g” that defines the section above
p1 ”. "_bs" indicating that the section is a bss section
s ", the section name is" gp1_bss ", and the area size is 4000. Also, for "b", the section is added to the global pointer "gp2" defined above. "_b" indicating that the section is a bss section
"ss", the section name is "gp2_bss", and the area size is 40000. This is shown in FIG. 10C for easy understanding. Similarly, for “c”, the section is assigned to the global pointer “gp1” defined above. "_bss" indicating that the section is a section name "gps"
1_bss ”, and the area size is 10000. FIG. 3D shows this for easy understanding.

Therefore, when the size of each section regarding gp1 in FIG. 9A is determined, “gp1_bs
For "s", the attribute "_g" is used as a global pointer.
p_gp1 ", the size is 40000,
“Gp1_data” has the attribute “_gp1” as a global pointer and a size of 128.

When the size of each section related to gp2 is determined, “gp2_bss” has the attribute “_gp2” as a global pointer, the size is 40000, and “gp2_data”
For the attribute “_gp” as a global pointer
2 "and the size is 128. further,. te
The size of the xt area is 400. In FIG. 10 (e),
This is shown for ease of understanding.

On the other hand, when the size of each section related to gp1 in FIG. 9B is determined, “gp1_bs
For "s", the attribute "_g" is used as a global pointer.
p1 ”and the size is 10,000. Also,
“Gp1_data” has the attribute “gp1” as a global pointer and a size of 64.
FIG. 3 (f) shows this for easy understanding. FIG. 5 shows the size of this result for each section. That is, the sum of the above-described FIGS. 10 (e) and (f) is shown.

Subsequently, in step S105, a plurality of sections relating to each global pointer are successively arranged, and the head address of the section is determined.
At 06, the value of each global pointer is determined. Thus, the initial value of the global pointer “gp1” is 0
x00008000 and the global pointer "gp
The initial value of “2” is 0x00014410.

Subsequently, in step S107, an address and an offset are generated based on the relocation information, and the relocation address is calculated. The calculation of the relocation address is shown in FIG.
As shown in (2), it is calculated by (label name)-(initial value of corresponding global pointer). In this step, after determining the addresses of all the labels and the initial values of the global pointers, an offset value is obtained in accordance with the calculation formula shown in FIG. 6, and the offset value is directly embedded in the output executable object.

FIG. 13 shows the memory image after the combined editing. For each assembler source, the initial value of each global pointer is assigned to gp1 and gp2. For gp1_bss and gp1_data, continuous areas are allocated, and gp1 is allocated at a position of 32 KB from the head. Similarly, gp2_bss
And gp2_data are allocated contiguous areas,
Gp2 is allocated to the position of 32 KB from the head. The instruction code is. It is organized into text sections. When executing these codes, it is necessary to set the value of the global pointer corresponding to each task in the gp register.

[0060] The gpsecon pseudo-instruction follows. . sbss and. Defines the section name to which the variable declared by the sdata pseudo-instruction is allocated. One. For the name given to the gpsecon pseudo-instruction, the data without an initial value is _bss
Is assigned to a section having a name concatenated with _data, and data with an initial value is assigned to a section having a name concatenated with _data. In the example of FIG. gpsect
After ion gp1,. Variable a declared in sbss
Contains in a section named gp1_bss,. s
The variable x declared in data is allocated to a section named gp1_data.

Note that the% gpof used in the present embodiment is
The fs macro has two arguments. The first argument is the label of the variable to be referenced, and the second argument is the name indicating the global pointer section to which the label belongs. As an instruction operand, it is necessary to write a register name used as a global pointer after the% gpoffs macro. It is the programmer's responsibility that the correct global pointer value be stored in the register when the instruction is executed.

In this embodiment, the description of the portion relating to the initialization of the gp register and the replacement of the gp register accompanying the task switching is omitted, but in the function of the normal multitask operating system, the gp register is used as a task-specific resource. Can be easily realized by assigning

As described above, according to the program combination editing method of the present embodiment, even if the task is divided into several parts and the size of the data to be handled does not fit in one global pointer section, the register is used as the global pointer. May be able to reduce the number of instructions. Such programs can be combined and edited.

Fourth Embodiment Next, a fourth embodiment will be described with reference to the drawings. This embodiment is different from the first and second embodiments in that the global pointer corresponding to the% gpoffs macro is the. % is determined depending on the context with the gpsection specification.
A function for automatically obtaining a variable from a label given as an argument to the gpoffs macro is added, and when a program similar to that of the third embodiment is created, only one argument of the% gpoffs macro is used. The flow of the combined editing process according to this embodiment will be described with reference to FIG. In the combination editing process, first, a relocatable object file to be input is read (step S201). The attribute of the section information is checked, and a list of global pointers is created (step S202).

Subsequently, for each section, an offset is determined for the label in the section (step S20).
3), the contents of the same section name are concatenated to determine the size of each section.
An area of the capacity specified in the object is secured (step S204). Subsequently, the start address of each section is determined (step S205). Subsequently, for each of the global pointer sections, the value of the global pointer for referring to it is determined (step S206). At this time, the capacity of the section
If the capacity exceeds the accessible capacity using one global pointer, the processing is stopped as an error and an error message is generated. Subsequently, for each of the global pointer sections, the offset of the variable belonging thereto from the value of the global pointer determined above is calculated, and a list is created (step S207). continue,
Address calculation and offset generation are performed based on the relocation information (step S208), where% gpoffs
The relocation information corresponding to the macro is processed by searching and referring to the label list generated above. Finally, an executable object is generated (Step S209).

Next, a specific example of this embodiment will be described with reference to the drawings. The difference between the flowchart of this embodiment shown in FIG. 14 and the flowchart used in the above-described first to third embodiments is that step 207 is added. Therefore, these steps will be described, and the other steps are as described above.
The description is omitted. Note that. The function of the gpsection pseudo-instruction is the same as that of the third embodiment.

In step S207, an offset for the global pointer is calculated. This calculation is performed by calculating the difference between the two addresses for a calculation formula written in the relocation information. FIG. 15 shows the result of this calculation.

FIG. 13 shows the memory image after the combined editing. The memory image after the combined editing is the same as in the third embodiment. However, the% gpoffs macro is 1
It differs from the third embodiment in that it has only one argument.
This argument is the label of the referenced variable. In response to this macro, the assembler generates relocation information indicating an offset value of the variable label from the global pointer, and outputs the information to the relocatable object file. This relocation information does not include information on the corresponding global pointer.

In the third embodiment, it is necessary to specify a global pointer corresponding to a label as an argument for the% gpoffs macro. However, according to the program combination editing method of this embodiment, this is not necessary. . As a result, the execution speed of the output executable object is improved.

A program for realizing the above-described program combination editing method can be stored in a recording medium. The recording medium can be read by a computer system, and the above-described program combination editing method can be realized while controlling the computer by executing the program. Here, the recording medium includes a device capable of recording a program, such as a memory device, a magnetic disk device, and an optical disk device.

[0071]

As described above, according to the program combination editing method and the machine readable recording medium recording the program combination editing program according to the present invention,
In the creation of a program to be combined and edited collectively, the number of instructions required for access can be reduced by allowing each task to have a unique global pointer.

[Brief description of the drawings]

FIG. 1 is a flowchart illustrating a program combination editing method according to first and second embodiments.

FIG. 2 is a diagram showing a source program of an assembler used in the first embodiment.

FIG. 3 is a table for explaining an operation of a program combination editing method according to the first and second embodiments.

FIG. 4 is a table showing section information.

FIG. 5 is a table showing a result of calculating an address based on relocation information;

FIG. 6 is a diagram showing a formula for calculating relocation information output in a relocatable object.

FIG. 7 is a diagram showing a memory image after combined editing.

FIG. 8 is a diagram showing a source program of an assembler used in a second embodiment.

FIG. 9 is a diagram showing an assembler source program used in the third and fourth embodiments.

FIG. 10 is a chart for explaining the operation of the program combination editing method according to the third and fourth embodiments.

FIG. 11 is a chart showing section information.

FIG. 12 is a table showing a result of calculating an address based on relocation information;

FIG. 13 is a diagram illustrating a memory image after combined editing according to the third and fourth embodiments.

FIG. 14 is a flowchart illustrating a program combination editing method according to the fourth embodiment.

FIG. 15 is a diagram illustrating a calculation result of a relocation address.

FIG. 16 is a table showing a calculation formula of relocation information output to a relocation-possible object.

FIG. 17 is a diagram for explaining an access operation of the MIPS architecture.

FIG. 18 is a diagram illustrating an access operation using a global pointer of the MIPS architecture.

FIG. 19 is a flowchart showing a conventional program combining / editing method.

FIG. 20 is a diagram showing an assembler source program used in a conventional program combination editing method.

FIG. 21 is a diagram for explaining an operation of a conventional example.

FIG. 22 is a diagram showing a memory image after combined editing according to a conventional example.

Claims (8)

[Claims] 1. A combined editing method of a program for performing combined editing on a plurality of assembler source programs including a global pointer section, wherein data without an initial value and data with an initial value are stored in the global pointer section. A program combination and editing method, wherein an executable object in which a global pointer unique to each of the global pointer sections included in the plurality of assembler source programs is provided is output individually for each global section. 2. A combined editing method of a program for performing combined editing on a plurality of assembler source programs including a global pointer section, wherein the global pointers of the plurality of global pointer sections specified in the assembler source program are deleted. A relative offset from the value is obtained, an address in each of the global pointer sections is calculated using the relative offset value, data without an initial value and data with an initial value are individually arranged for each global pointer section, A program combination editing method, comprising outputting an executable object in which a unique global pointer is provided for each global pointer section included in the plurality of assembler source programs. 3. A combined editing method for a program for performing combined editing on an assembler source program including a plurality of global pointer sections, wherein the global pointers related to the plurality of global pointer sections specified in the assembler source program. A relative offset from the value is obtained, an address in each of the global pointer sections is calculated using the relative offset value, data without an initial value and data with an initial value are individually arranged for each global pointer section, A program combination editing method, comprising outputting an executable object in which a unique global pointer is provided for each global pointer section included in the plurality of assembler source programs. 4. A method for combining and editing a plurality of assembler source programs including a global pointer section, the plurality of assembler source programs comprising a plurality of global pointer sections specified in the assembler source program. The relative offset from the value of the pointer is obtained, the address in each of the global pointer sections is calculated using the relative offset value, and data without an initial value and data with an initial value are individually arranged for each global pointer section. And outputting an executable object provided with a unique global pointer in each global pointer section included in the plurality of assembler source programs. 5. A combined editing method for combining and editing a plurality of assembler source programs including a global pointer section, wherein a label in the global pointer section is specified in the assembler source program. Calculating an offset from the value of the global pointer relating to the plurality of global pointer sections; data without an initial value and data with an initial value are individually arranged for each global pointer section; And outputting an executable object in which a global pointer unique to the global pointer section is provided. 6. A method for combining and editing a plurality of assembler source programs including a global pointer section, wherein the assembler program is input, and a global pointer used in the global pointer section is input. Create a list, determine an offset for a label in the section for each section of the global pointer section, determine a size of each section for each section, and determine a plurality of sections related to each global pointer. Place them consecutively, determine the start address of the section, determine the value of each global pointer, generate the address and offset based on the relocation information,
A method for editing a program, comprising calculating a relocation address and outputting a generated executable object. 7. A machine-readable recording medium recording a combined editing program of a program for performing combined editing on a plurality of assembler source programs including a global pointer section, wherein an initial value is set for the global pointer section. None data and data with an initial value are individually arranged for each global section, and an executable object provided with a unique global pointer for each global pointer section included in the plurality of assembler source programs is output. A machine-readable recording medium on which a program combination editing program is recorded. 8. A machine-readable recording medium recording a combined editing program of a program for performing combined editing on a plurality of assembler source programs including a global pointer section, wherein the assembler program is input; Creating a list of global pointers used in the global pointer section; determining, for each section of the global pointer section, an offset with respect to a label within the section; Determining the size; arranging a plurality of sections related to each global pointer in succession; determining the head address of the section; determining the value of each global pointer; To generate a dress and offset,
A machine-readable recording medium recording a program editing program, comprising: calculating a relocation address; and outputting the generated executable object.