JP5406549B2 - Program conversion method and apparatus - Google Patents

Description

  The present invention relates to a program conversion method and apparatus for converting a source code of a given program.

  Conventionally, in the program conversion technique, by giving a performance index to a specified section of a program, the section is automatically set as a unit for scheduling a task (that is, OS (operating system; the same applies hereinafter)). The proposal for the purpose of providing a program conversion method and apparatus that add a performance index by dividing the data into (1) is known. The outline of the proposal is that a section and index acquisition means acquires a section code representing a section embedded in the program and performance index information embedded in the program in association with the section code, and the task code conversion means The section code acquired by the section and the index acquisition means is separated as a task code, and a code indicating the task head and a code indicating the task end are added. The task index adding means adds a performance index to be input to the scheduler to the task (see, for example, Patent Document 1). Further, as an additional function to the scheduler of the OS, a technique related to a mechanism that makes it possible to ensure the performance of a CPU resource to be allocated to a task is also known (see, for example, Non-Patent Document 1). ).

JP 2006-126977 S. Oikawa and R. Rajkumar, “Portable RK: A Portable Resource Kernel for Guaranteed and Enforced Timing Behavior”, Proceedings of the Fifth IEEE Real-Time Technology and Applications Symposium, 1999.

  Under a multitasking environment, processing related to each task is executed in a state where a plurality of tasks share the same CPU. That is, by dividing the CPU processing capacity in time and allocating CPU resources to each task, it becomes possible to execute the plurality of tasks simultaneously. Generally, a scheduler function for allocating CPU resources to each task is provided as an OS function. In a multitasking environment where the same CPU is shared by multiple tasks, execution is the real time from the start of execution to the end of execution for the entire program or a certain section of the program The time is longer than the calculation time, which is the time that the program used the CPU from the start of execution to the end of execution in the section. The amount of CPU resources allocated to each task by the scheduler depends on the environment at the time of execution of the program (because processing of other tasks is interrupted), so it has the same calculation time or the same calculation amount. Even so, the execution time of the program varies depending on the situation.

  In particular, in the case of an embedded system such as a digital broadcast receiver, real-time performance of the program is required. The real-time property of a program is a property related to the execution time of the program, for example, a property that the execution time in a specific section in the program falls within a predetermined time. In a video display processing device such as the digital broadcast receiver described above, if the execution of a program for video display cannot be completed within a certain time, the output video from the video display processing device is disturbed. It is important to maintain the real-time nature of the program.

  However, in a situation where a plurality of tasks as described above are operating with the same CPU being shared, as described above, the CPU resource allocated by the scheduler depends on the environment when the program is executed. Because of this, it is difficult to maintain the real-time nature of the program. Therefore, a technique has been proposed for ensuring the CPU resources allocated to each task by giving a performance index to the program in the form of CPU resource utilization. This is the technique disclosed as Patent Document 1 described above and the technique disclosed as Non-Patent Document 1.

  By the way, when the technology disclosed in Patent Document 1 and Non-Patent Document 1 described above is adopted, the program is provided in a form of a utilization rate of CPU resources allocated for executing the program. Although performance can be ensured, there is a problem that the performance of the program cannot be maintained when the performance index is given in the form of execution time for the program or a specific section in the program. Arise. In addition, when the technique related to securing the CPU resource utilization rate included in the conventional technique is adopted to maintain the performance of the program related to the execution time described above, the amount of the secured CPU resource is set to an appropriate CPU. If the resource utilization rate is not satisfied, there may be a situation in which the performance index related to the designated execution time for the program cannot be maintained. On the other hand, if the amount of reserved CPU resources is excessive than the appropriate CPU resource utilization rate, the amount of CPU resources allocated to tasks other than the task decreases, and as a result, the responsiveness of other tasks decreases. There is a possibility to provoke.

  It is known that the amount of calculation of a program varies depending on the execution time of the same program, depending on the internal state of the program, various data input to the embedded system from the outside, and the like. When the calculation amount of the program fluctuates as described above, a specific section in the program must be executed within the execution time given as a performance index. Therefore, a CPU suitable for executing the specific section Since the resource amount also fluctuates, it becomes difficult to obtain an appropriate CPU resource utilization rate in advance as described above.

  Accordingly, an object of the present invention is to maintain a performance index related to a specified execution time for a program in a multitasking environment, and to prevent a specific task from occupying excessive CPU resources. The purpose is to automatically generate such a performance ensuring program.

  The program conversion apparatus according to the first aspect of the present invention converts a source code of a given program, and is provided with a performance index of the program from one place or a plurality of places in a certain section of the program. The program is started from the execution time calculation unit that calculates the maximum time required for execution until the end of the section, the maximum execution time calculated by the execution time calculation unit, and the performance index. A function adding unit that gives a function for obtaining a CPU resource ratio to be secured for execution from the one or more places to the end of the section in the CPU to be executed, and the function addition The source code of the program to which the function for determining the CPU resource ratio is given by the section is allocated to the program that secures the calculated CPU resource ratio. Comprising a program conversion unit that converts the grams, the.

In a preferred embodiment according to the first aspect of the present invention, in certain the section of the program performance index of the program is given, for detecting whether the cause on the variation in the calculated amount of the program has occurred A variation cause detection unit is further provided, and one or a plurality of locations in a certain section of the program is a position where the cause of the variation detected by the variation cause detection unit exists.

  In an embodiment different from the above, the cause of the fluctuation is a branch destination of a conditional branch existing in the section.

  In another embodiment different from the above, the cause of the fluctuation is a loop existing in the section, and the one place or a plurality of places are after the determined number of loops of the loop in the section. This part is included.

  In another embodiment different from the above, the execution time calculation unit obtains the maximum value of the execution time according to the determined number of loops existing in the section detected by the variation cause detection unit.

In another embodiment different from the above, the execution time calculation unit is generated by the generation unit that generates a profile acquisition code for obtaining the time required to execute the section of the program, and generated by the generation unit. An execution unit that executes a profile acquisition code; a calculation unit that calculates a maximum value of the execution time from profile data acquired by executing the profile acquisition code by the execution unit;
Is provided.

  Further, in another embodiment different from the above, the performance index is a time required for executing the section of the program.

  The program conversion method according to the second aspect of the present invention converts the source code of a given program, from one place or a plurality of places in a certain section of the program to which the performance index of the program is given. The program is started from the first step of calculating the maximum time required for execution until the end of the section, the maximum value of the execution time calculated in the first step, and the performance index. A second step of adding to the source code of the program a function for obtaining a CPU resource ratio to be secured for execution from the one or more locations to the end of the section in the CPU The source code of the program to which the function for obtaining the CPU resource ratio in step 2 is assigned is obtained as the CPU resource ratio obtained above. And a third step of converting the program to be secured.

  According to the present invention, it is possible to maintain a performance index related to a specified execution time for a program in a multitasking environment, and to prevent a specific task from occupying excessive CPU resources. It is possible to automatically generate a performance guarantee program.

The functional block diagram which shows an example of the whole structure of the system in which the program conversion apparatus which concerns on one Embodiment of this invention is integrated. A process until a performance ensuring program is generated through each functional block from a functional block showing an example of an internal configuration of the program conversion apparatus shown in FIG. 1, a program source, a section and an execution time request designation code; Explanatory drawing which showed. FIG. 3 is an explanatory diagram showing an example of a program source including a conditional branch that is a kind of the program source shown in FIG. 2. FIG. 3 is an explanatory diagram showing an example of a program source including a loop that is a kind of the program source shown in FIG. 2. Explanatory drawing which showed an example of the area and execution time request | requirement designation | designated code shown in FIG. The flowchart which shows an example of the processing operation performed in the calculation amount fluctuation | variation factor extraction part shown in FIG. FIG. 7 is an explanatory diagram illustrating an example of a branch destination list by a conditional branch generated by a calculation amount variation factor extraction unit in the operation of the conditional branch extraction process illustrated in FIG. 6. The flowchart which shows the detail of operation | movement of the conditional branch extraction process shown in FIG. 6 by the calculation amount fluctuation | variation factor extraction part shown in FIG. FIG. 7 is an explanatory diagram showing an example of a loop list generated by a calculation amount variation factor extraction unit in the operation of the loop extraction process shown in FIG. 6. The flowchart which shows the detail of operation | movement of the loop extraction process shown in FIG. 6 by the calculation amount fluctuation | variation factor extraction part shown in FIG. The flowchart which shows an example of the processing operation performed in the relationship derivation | leading-out part of the variation factor and calculation amount shown in FIG. FIG. 3 is an explanatory diagram illustrating an example of a profile acquisition code generated in the operation of the relationship derivation process between the variation factor and the calculation amount by the relationship deriving unit between the variation factor and the calculation amount illustrated in FIG. FIG. 12 is a flowchart showing details of an operation of a relationship derivation process between the variation factor and the calculation amount shown in FIG. An example of a table showing the correspondence between the calculation time measurement position generated in the function approximation processing of the calculation time shown in FIG. 11 and the remaining section maximum calculation time by the relationship deriving unit between the variation factor and the calculation amount shown in FIG. FIG. An example of a table showing the correspondence between the loop count value generated in the function approximation processing of the calculation time shown in FIG. 11 and the remaining section maximum calculation time by the relationship deriving unit between the variation factor and the calculation amount shown in FIG. FIG. The graph approximated by the linear function (primary formula) which shows the relationship between the loop frequency value shown in FIG. 15, and remaining section maximum calculation time. 12 is a flowchart showing details of the operation of the function approximation process of the calculation time shown in FIG. 11 by the relationship deriving unit between the variation factor and the calculation amount shown in FIG. 2. The flowchart which shows an example of the processing operation performed in the required CPU resource amount calculation function addition part shown in FIG. The flowchart which shows an example of the processing content regarding the required CPU resource amount estimation code performed in the required CPU resource amount calculation function addition part shown in FIG. The flowchart which shows an example of the processing operation performed in the required CPU resource amount calculation function addition part shown in FIG. FIG. 5 is an explanatory diagram showing an example of a performance ensuring program that the program conversion apparatus shown in FIG. 2 outputs based on the program source shown in FIG. 4.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

  FIG. 1 is a functional block diagram illustrating an example of the overall configuration of a system in which a program conversion apparatus according to an embodiment of the present invention is incorporated (hereinafter, referred to as “embedded system”). In the following description, a digital broadcast receiving system is taken as an example of an embedded system.

  As shown in FIG. 1, the digital broadcast receiving system 100 includes a digital broadcast receiver main body (hereinafter referred to as “receiver main body”) 1 and a remote controller (hereinafter referred to as “remote controller”) 3. And an antenna 5, a display device 7, and a program conversion device 9. The receiver main body 1 includes a storage device 11, a processor 13, a signal receiving unit 15, and a digital broadcast tuner 17.

  The remote control 3 is provided with an operation unit (not shown) in which a plurality of keys (not shown) for sending various operation command signals to the receiver body 1 are arranged. When an operation command signal is generated in the remote controller 3 by the user operating one of the plurality of keys (not shown), the remote controller 3 receives, for example, infrared light on which the operation command signal is superimposed. The operation command signal is transmitted to the receiver body 1 by radiating toward the machine body 1.

  The antenna 5 outputs an electrical signal generated due to reception of a digital broadcast signal wave transmitted from a broadcast station (not shown) to the receiver body 1. The display device 7 displays the image information as a visible image according to the display output from the receiver body 1 side.

  In the receiver main body 1, the storage device 11 is subjected to processing that is converted into a performance ensuring program in the program conversion device 9 among various programs included in the digital broadcast receiver software 19 under the control of the processor 13. The program (hereinafter referred to as “conversion program”) and the program that has not been subjected to the conversion process are stored together. The digital broadcast receiver software 19 includes a remote controller task (hereinafter referred to as “remote control task”) 21, a tuner task 23, a video decode task 25, an audio decode task 27, and a data broadcast browser task. 29.

  In the digital broadcast receiver software 19, the remote control task 21 has a function of executing reception processing of the operation command (signal) corresponding to the operation command (signal) based on the operation of the remote control 3 by the user. There are a tuner task 23, a video decoding task 25, an audio decoding task 27, and a data broadcast browser task 29 as programs for individually executing appropriate response processing for the reception processing.

  That is, the tuner task 23 is activated in accordance with a channel selection operation in the digital broadcast tuner 17 (of the receiver main body 1) based on a channel selection command (via the operation of the remote controller 3) by the user, and performs a received program switching process. Has the function to execute. The video decoding task 25 has a function of executing a decoding process of video information included in a digital broadcast signal wave received by the tuner task 23 via the digital broadcast tuner 17 when activated. The audio decoding task 27 has a function of executing a decoding process of audio information included in a digital broadcast signal wave received by the tuner task 23 via the digital broadcast tuner 17 when activated. The data broadcast browser task 29 has a function of acquiring data broadcast content (from the digital broadcast tuner 17) by activation and a function of performing a decoding process on the acquired data broadcast content. Of the above programs, the tuner task 23, the video decoding task 25, and the audio decoding task 27 correspond to the conversion program, and the remaining remote control task 21 and the data broadcasting browser task 29 are transferred to the performance ensuring program. This program is not subjected to the conversion process.

  Of the programs stored in the storage device 11 as the performance ensuring program, the video decoding task 25 and the audio decoding task 27 are required to output image information and audio information without interruption. There is a time restriction that each decoding process must be completed within a time determined by a frame rate (of image information). As for the tuner task 23 among the programs stored in the storage device 11 as the performance ensuring program, the receiver body 1 obtains image information and audio information from the digital broadcast signal wave received via the antenna 5. A restriction is imposed on the time for executing the extraction process. Therefore, in the present embodiment, among the various programs included in the digital broadcast receiver software 19, the tuner task 23, the video decoding task 25, and the audio decoding task 27, which are tasks having real-time properties, are programs. After being converted into a performance ensuring program in the conversion device 9, it is stored in the storage device 11.

  On the other hand, for the remote control task 21 and the data broadcasting browser task 29, there is no clear restriction that the processing operation to be executed within a predetermined time must be completed. Unlike the decoding task 25 and the audio decoding task 27, the data is stored in the storage device 11 as it is. In addition, as described above, when a long time is required for execution of the processing operation even if there is no clear restriction, there is a problem that the responsiveness in the receiver main body 1 is lowered or the operability is lowered. There is also a risk of it occurring.

  The functions of the various programs (21, 23, 25, 27, and 29) included in the digital broadcast receiver software 19 described above are executed by the processor 13.

  In addition to the above, the storage device 11 manages various (hardware) resources such as the receiver main body 1 and an OS for allowing an application (program) to use the various (hardware) resources. And applications (programs) are stored.

  The signal receiving unit 15 receives infrared light transmitted from the remote controller 3 on which the operation command signal is superimposed under the control of the processor 13 and performs predetermined signal processing on the operation command signal. Output to the processor 13. The digital broadcast tuner 17 controls digital broadcast signal waves (video information and audio information) received from the plurality of broadcast stations received by the antenna 5 in accordance with a channel selection command given by the user via the remote controller 3 under the control of the processor 13. In addition, a channel selection processing operation for extracting a digital broadcast signal wave from a broadcast station desired by the user is executed. The digital broadcast signal wave desired by the user extracted by performing the channel selection process in the digital broadcast tuner 17 is output to a signal processing system (not shown) that performs signal processing such as demodulation under the control of the processor 13. Is done.

  The processor 13 puts each part constituting the receiver main body 1 such as the storage device 11, the signal receiving unit 15, and the digital broadcast tuner 17 under the control, and manages the processing operation of each unit. That is, the processor 13 performs the processing based on various operation command signals given from the user via the remote controller 3, the remote control task 21, the tuner task 23, the video decoding task 25 stored in the storage device 11, Corresponding programs are activated from the audio decoding task 27, the data broadcasting browser task 29, and the like. Processing operations executed by the digital broadcast receiver software 19 such as the remote control task 21, the tuner task 23, the video decode task 25, the audio decode task 27, and the data broadcast browser task 29 using the hardware resources of the processor 13, respectively. Since the details have been already described, repeated description thereof is omitted here.

  As described above, the program conversion device 9 uses the performance ensuring program for the tuner task 23, the video decoding task 25, and the audio decoding task 27 in the digital broadcast receiver software 19 to be stored in the storage device 11. Is converted into a conversion program. The program conversion device 9 stores the tuner task 23, the video decoding task 25, and the audio decoding task 27 after being converted into the performance ensuring program in the storage device 11 through the processor 13 as the digital broadcast receiver software 19. The program conversion device 9 will be described in detail below.

  FIG. 2 shows the above-described function block, an example of the internal configuration of the program conversion apparatus 9 shown in FIG. 1, a program source, a section and an execution time request designation code (hereinafter referred to as “code”). It is explanatory drawing which showed the process until a performance ensuring program is produced | generated through each functional block.

  As shown in FIG. 2, the program conversion apparatus 9 inputs a program source 39 and a code 41, and performs a predetermined conversion process on the program source 39 to output a performance ensuring program 43. The program conversion apparatus 9 includes a calculation amount variation factor extraction unit (hereinafter referred to as “extraction unit”) 31 and a relationship derivation unit (hereinafter referred to as “derivation unit”) between the variation factor and the calculation amount. 33, a required CPU resource amount calculation function addition unit (hereinafter referred to as “calculation function addition unit”) 35, and a CPU resource allocation function addition unit (hereinafter referred to as “allocation function addition unit”) 37. .

  The extraction unit 31 inputs a program source that includes conditional branches of various aspects and a program source that includes loops of various aspects as the program source 39, and outputs a plurality of types of sections and execution time requests to a code (that is, (Section and execution time request designation code) 41. The extraction unit 31 executes conditional branch extraction processing and loop extraction processing for the section of the input program source 39 in which the execution time request is specified by the code 41. The program source 39 reflecting the execution result of the conditional branch extraction process and the loop extraction process is output from the extraction unit 31 to the derivation unit 33.

  When the derivation unit 33 inputs the program source 39 reflecting the execution result of the conditional branch extraction process and the loop extraction process from the extraction unit 31, the derivation unit 33 generates a profile acquisition code generation process, a calculation amount for the program source 39. Profile acquisition processing and execution time function approximation processing are performed. The program source 39 reflecting the execution results of the profile acquisition code generation process, the calculation amount profile acquisition process, and the execution time function approximation process is output from the derivation unit 35 to the calculation function addition unit 35.

  The calculation function adding unit 35 receives the program source 39 reflecting the execution results of the profile acquisition code generation process, the calculation amount profile acquisition process, and the execution time function approximation process from the derivation unit 35, and A process of inserting the required CPU resource amount estimation code and the required CPU resource amount re-estimation code is executed for the source 39 by a predetermined method. The program source 39 reflecting the execution result of the processing is output from the calculation function addition unit 35 to the assignment function addition unit 37.

  The allocation function adding unit 37 receives the required CPU resource amount estimation code or the program source 39 reflecting the execution result of the processing for inserting the required CPU resource amount re-estimation code from the calculation function adding unit 35, and then Based on the CPU resource amount estimation code and the necessary CPU resource amount re-estimation code, predetermined arithmetic processing is executed. As a result, the program source 39 is program-converted, and the program source (39) that has undergone the program conversion processing operation is sent from the program converter 9 to the receiver main body 1 shown in FIG. Will be output.

  FIG. 3 is an explanatory diagram showing an example of a program source including a conditional branch which is a kind of the program source 39 shown in FIG.

  In the program source 45 shown in FIG. 3, reference numeral 45 a is an item corresponding to the line number “5” in the program source 45, and indicates a conditional branch in the execution time request designation section designated by the code 41 described above. Show. Reference numeral 45b is an item corresponding to the line number “6” in the program source 45, and indicates a branch destination when a predetermined condition relating to the conditional branch 45a is satisfied. Reference numeral 45c denotes an item corresponding to the line number “9” in the program source 45, and indicates a branch destination when a predetermined condition related to the conditional branch 45a is not satisfied. Note that “src_a.c” described in the upper left part of FIG. 3 indicates the file name of the program source 45 shown in FIG.

  FIG. 4 is an explanatory diagram showing an example of a program source including a loop which is a kind of the program source 39 shown in FIG.

  In the program source 47 shown in FIG. 4, reference numeral 47 b is an item corresponding to the line number “7” to the line number “10” in the program source 47, and is an execution time request designation designated by the code 41 described above. Indicates a loop in a section. Reference numeral 47 a indicates an item corresponding to the line number “5” in the program source 47. The number of loops related to the loop 47b is either a constant or a variable (n). Reference numeral 47a denotes the variable (n) when the number of loops related to the loop 47b is a variable (n). The position, that is, the position of the process immediately after the loop 47b is determined, in other words, the position before the loop 47b and the place where the value of “n” is updated last. Note that “src_b.c” described in the upper left part of FIG. 4 indicates the file name of the program source 47 shown in FIG.

  FIG. 5 is an explanatory diagram showing an example of the code (that is, the section and execution time request designation code) 41 shown in FIG.

  In FIG. 5, the code 41 includes information about a specific section for each program source (39) identified by a file name, information regarding a start point, information regarding an end point, and information regarding the specific section within a limited time. And information related to the requested execution time when requesting execution of processing. Incidentally, in the example shown in FIG. 5, in the section where the start point in the program source (39) of the file “src_a.c” is defined by the line number “2” and the end point is defined by the line number “16”, It is required to execute the process in “10 milliseconds”. In the section where the start point in the program source (39) of the file “src_b.c” is defined by the line number “3” and the end point is defined by the line number “13”, the processing in the section is performed in “20 milliseconds”. It is required to execute.

  FIG. 6 is a flowchart showing an example of processing operations executed in the extraction unit 31 shown in FIG.

  In FIG. 6, when the extraction unit 31 inputs the program source 39 and the code 41, the extraction source 31 in the section of the program source 39 specified by the code 41 to execute the processing operation within a predetermined time. First, the conditional branch extraction processing operation is executed (step S51), and then the loop extraction processing operation is executed (step S52). Thereby, the series of processing operations shown in FIG. 6 is completed.

  FIG. 7 shows an example of a branch destination list (hereinafter referred to as “branch destination list”) generated by the conditional branch generated by the extraction unit 31 in the operation of the conditional branch extraction process shown in FIG. 6 (step S51). It is explanatory drawing shown.

In FIG. 7, the branch destination list 53 is composed of a plurality of file name recording fields 55 and a plurality of line number recording fields 57 associated with each of the plurality of file name recording fields 53.
Each file name recording column 53 has a file name (information) for identifying the individual program source 39 that is a processing target of the program conversion device 9, and each line number recording column 55 has a corresponding name. In each program source (39), the line number (information) indicating the branch destination searched by the extraction unit 31 is recorded. The example shown in FIG. 7 relates to the program source 39 having the file name “src_a.c”, and the condition relating to the conditional branch (45a) searched by the extraction unit 31 is established for the line number “6”. This indicates that the extraction unit 31 records the branch number as the branch destination and the line number “9” as the branch destination when the condition related to the conditional branch (45a) is not satisfied.

  FIG. 8 is a flowchart showing details of the conditional branch extraction processing operation (step S51) shown in FIG. 6 performed by the extraction unit 31 shown in FIG.

  In FIG. 8, the extraction unit 31 first inputs the code 41, and then inputs the section of the program source 39 designated by the code 41 (step S61). Next, the extraction unit 31 searches whether a conditional branch exists in the section. For example, if the program source 39 is the program source 45 with the file name “src_a.c” shown in FIG. 3, there is a conditional branch (reference 45a) (YES in step S62). In this case, if the condition related to the conditional branch 45a searched by the extraction unit 31 is satisfied, the branch destination is the position of the line number “6” indicated by the reference numeral 45b. On the other hand, if the condition related to the conditional branch 45a is not satisfied, the branch destination is the position of the line number “9” indicated by the reference numeral 45b.

  If “YES” in the step S62, the extraction unit 31 sets the line number “6” as the branch destination when the condition related to the conditional branch 45a is satisfied, and if the condition related to the conditional branch 45a is not satisfied. Records the line number “9” as the branch destination in the branch destination list 53 shown in FIG. 7 (step S63), and returns to the processing operation shown in step S62. On the other hand, if the conditional branch cannot be found in the section specified by the code 41 in the program source 39 to be processed as a result of the search by the extraction unit 31 (NO in step S62), the extraction unit 31 terminates a series of processing operations related to the conditional branch extraction processing shown in FIG.

  FIG. 9 is an explanatory diagram showing an example of a loop list generated by the extraction unit 31 in the operation of the loop extraction process shown in FIG. 6 (step S52).

  In FIG. 9, the loop list 65 includes a loop count variable (the loop count has two types, a constant and a variable, but here, for convenience of explanation, only a variable is taken as an example) a recording column 67, The process position record field 69 (hereinafter referred to as “process position record field”) 69 immediately after the loop count is determined, and the process position record field 69 includes a file name record field 71 and a line number record. It has a column 73.

In the loop count variable recording column 67, the loop count indicated by the variable “n” retrieved by the extraction unit 31 in the corresponding individual program source (39) is the file name recording column in the processing position recording column 69. In 71, file names (information) for identifying individual program sources 39, which are processing targets of the program conversion apparatus 9, are recorded. Further, the line number recording column 73 in the processing position recording column 69 contains the number of loops indicated by the loop searched by the extraction unit 31 in the corresponding individual program source (39), that is, the variable “n”. The position at which the process immediately after the determination is executed, in other words, the number of the line next to the line where the value of the variable “n” is updated last before the loop is recorded. The example shown in FIG. 9 relates to the program source 39 with the file name “src_b.c”, and the line number “6” is the loop number variable “n” of the loop 47 b searched by the extraction unit 31.
This is the position where the process immediately after the determination is executed.

  FIG. 10 is a flowchart showing details of the operation (step S52) of the loop extraction process shown in FIG. 6 by the extraction unit 31 shown in FIG.

  In FIG. 10, the extraction unit 31 first inputs the code 41, and then inputs the section of the program source 39 designated by the code 41 (step S81). Next, the extraction unit 31 searches for a loop in the section. For example, if the program source 39 is the program source 47 having the file name “src_b.c” shown in FIG. 4, a loop exists (reference 47b) (YES in step S82). As apparent from FIG. 4, in the case of the program source 39 having the file name “src_b.c”, the loop count is a variable “n” (YES in step S83). Therefore, the extraction unit 31 records the variable “n” as the loop count in the loop count variable recording column 67 of the loop list 65 illustrated in FIG. 9 and the file name of the processing position recording column 69 illustrated in FIG. “Src_b.c” is recorded in the recording field 71, and the line number “6” is recorded in the line number recording field 73 as the processing position (step S84). Then, the processing returns to the processing operation shown in step S82. Even when the number of loops in the specified section described above of the program source 39 to be processed is a constant (YES in step S83), the processing operation shown in step S82 is resumed.

  If the loop is not searched as a result of the search in the section by the extraction unit 31, it is assumed that there is no loop (NO in step S82), and the series of processing operations shown in FIG.

  FIG. 11 is a flowchart illustrating an example of a processing operation executed in the derivation unit 33 illustrated in FIG.

  In FIG. 11, the deriving unit 33 inputs the program source 39, the branch destination list 53 and the loop list 65 output from the program extracting unit 31, respectively. Then, based on the branch destination list 53 and the loop list 65, a profile acquisition code generation process for generating a profile acquisition code is executed by inserting a code for profile measurement into the program source 39. (Step S91). Next, the generated profile acquisition code is executed under various conditions, and the calculation time related to the execution is recorded (step S92). Then, function approximation processing is executed for the recorded calculation time (step S93). As a result, the series of processing operations shown in FIG. 11 ends.

  FIG. 12 is an explanatory diagram illustrating an example of a profile acquisition code generated in the operation of the relationship derivation process between the variation factor and the calculation amount by the derivation unit 33 illustrated in FIG. FIG. 12 shows an example of profile acquisition code in the program source 39 whose file name is “src_b.c”. In the following description, in the processing operation in which the profile acquisition code is generated, the position where the code is inserted in the program source 39 (the start position of the specified section described above, the position recorded in the branch destination list 53, or the loop) The position recorded in the list 65 may be expressed as “calculation time measurement position”.

  In the profile acquisition code 95 shown in FIG. 12, the position immediately before the position indicated by the line number “6”, which is the position recorded in the loop list 65 (shown in FIG. 9), and the code described above. 41 (shown in FIG. 2), a request for execution of the processing operation within a predetermined time is set to the position immediately before the position indicated by the line number “3”, which is the start position of the designated section. A calculation time acquisition code from the position to the end of the section is inserted. That is, the position of the reference numeral 95c, which is the position immediately before the position of the line number "6", and the position of the reference numeral 95a, which is the position immediately before the position of the line number "3", are used in the profile acquisition described in detail later. In addition, codes for recording the current time obtained on the premise of running as a single task are inserted. In addition, a code for recording the current time obtained at the position is inserted at the position of the code 95d, which is the position immediately after the position of the line number “12”, and the current time 95d and the current time 95a. And a code for recording a value indicating a difference between the current time 95d and the current time 95c (represented by a reference numeral 95f) are inserted respectively. ing.

  Further, a code for recording the value of “n” (indicated by reference numeral 95b), which is a loop number variable, is inserted at a position immediately before the position of the row number “6” described above.

  FIG. 13 is a flowchart showing details of the operation of the relationship derivation process between the variation factor and the calculation amount shown in FIG. 11 by the derivation unit 33 shown in FIG. FIG. 13 shows details of the operation of the profile acquisition code generation process included in the processing operation shown in FIG.

  In the operation of the profile acquisition code generation process shown in FIG. 13, a list (that is, a branch destination list) relating to each calculation amount variation factor recorded in the extraction unit 31 with respect to the program source 39 input to the program conversion device 9. 53 and a loop list 65), a profile acquisition code is generated by inserting a code for profile measurement (for example, in the manner shown in FIG. 12).

  In FIG. 13, the derivation unit 33 performs code insertion processing as shown in step S102 and subsequent steps for each of the above-described calculation amount variation factors (step S101). In other words, the deriving unit 33 first checks whether the type of calculation amount variation factor described above is due to a loop (step S102). If it is determined as a result of the check (YES in step S102), the position immediately before “the position of the process immediately after determining the loop count” recorded in the loop list 65 described above (that is, the loop count). A code for recording the value of the loop number variable at that time is inserted at a position immediately after the position is determined (in the example of FIG. 12, the code indicated by reference numeral 95b corresponds) (step S103). When the processing operation is completed, the derivation unit 33 is then recorded in the list (either the branch destination list 53 or the loop list 65) regardless of the type of the (calculation amount) variation factor. For the position in the program source 39 (the position where the number of loops is determined or the branch destination position), the specified section described above from that position (that is, the section in which the request for executing the processing operation within the predetermined time is specified). A code for acquiring the calculation time of the section of the program source 39 until the end of the program source 39 is inserted (in the example of FIG. 12, the codes respectively indicated by reference numerals 95e and 95f correspond) (step S104). If the deriving unit 33 determines in step S102 that the type of calculation amount variation factor described above is not due to a loop (NO in step S102), the processing operation immediately proceeds to step S104. .

  When the processing operation shown in step S104 is completed, the deriving unit 33 starts the position (in the program source 39 in FIG. 12) of the section (that is, a section in which a request for executing the processing operation within a predetermined time is specified). Also, a code for acquiring the calculation time (in FIG. 12, the code at the position indicated by reference numeral 95a) is inserted. As a result, the calculation time of the entire (designated) section corresponding to the above request can be acquired (step S105). Then, the series of processing operations shown in FIG. 13 ends.

  As a method for acquiring the calculation time from the code insertion position described above in the program source 39 to the end of the section, on the premise that a single task is run when acquiring the profile, the code insertion position There is a method in which a code for recording the current time is inserted at a position where a section in which a request for executing a processing operation within a specified time ends, and a difference between the recorded times is obtained. Further, when CPU time information allocated to each task can be acquired from the scheduler, the information may be used.

  FIG. 14 is a table showing the correspondence between the calculation time measurement position generated by the derivation unit 33 shown in FIG. 2 and the calculation time function approximation process shown in FIG. It is described as an example of “correspondence table”.

  In FIG. 14, the correspondence table 107 includes line number information indicating the calculation time measurement position for each program source (39) identified by the file name, and the calculation in the remaining section from the calculation time measurement position (measurement start position). And information (unit milliseconds) related to the maximum value of time (that is, maximum calculation time). Incidentally, in the example shown in FIG. 14, the maximum calculation time of the remaining section whose calculation time measurement position is the line number “2” in the program source (39) with the file name “src_a.c” is “7.5 milliseconds”. The maximum calculation time of the remaining section whose calculation time measurement position in the program source 39 is the line number “6” is “2.3 milliseconds”. Further, the maximum calculation time of the remaining section whose calculation time measurement position is the line number “9” in the program source 39 is “4.2 milliseconds”. In addition, the maximum calculation time of the remaining section whose calculation time measurement position is the line number “3” in the program source (39) with the file name “src_b.c” is “14.2 milliseconds”.

  FIG. 15 is a table showing the correspondence between the loop count value generated in the function approximation process of the calculation time shown in FIG. 11 by the deriving unit 33 shown in FIG. It is described as an example.).

The correspondence table 109 shown in FIG. 15 relates to the program source 39 having the file name “src_b.c” already described. In the program source 39, the number of loops existing in the specified section described above is the variable “n”, and the processing position immediately after the determination of the number of loops is as shown in FIG. 6 "
Is the position. In the position immediately before the line number “6”, a code for recording the value of “n” which is a loop count variable (indicated by reference numeral 95b) is inserted.

  In the example shown in FIG. 15, the remaining section maximum calculation time when the loop count value is “0” is “5.6 milliseconds”, and the remaining section maximum calculation time when the loop count value is “1” is The remaining section maximum calculation time when the loop count value is “2” is “7.3 milliseconds”, and the remaining section maximum calculation when the loop count value is “3”. The time is “8.7 milliseconds”. Further, the remaining section maximum calculation time when the count value is “4” is “9.3 milliseconds”, and the remaining section maximum calculation time when the loop count value is “5” is “10.7 milliseconds”. Yes, the remaining section maximum calculation time when the loop count value is “6” is “12.5 milliseconds”, and the remaining section maximum calculation time when the loop count value is “7” is “12.7 milliseconds”. ".

  FIG. 16 is a graph approximated by a linear function (linear expression) showing the relationship between the loop count value shown in FIG. 15 and the remaining section maximum calculation time.

  In the graph shown in FIG. 16, the remaining section maximum calculation time (unit: millisecond) is set on the vertical axis, and the loop count variable value “n” is set on the horizontal axis. In FIG. 16, a line segment 111 represents a linear function y = 1.1x + 5.9. A broken line 113 connects points on the graph indicating the remaining section maximum calculation time corresponding to each loop number value when the loop number variable value “n” takes a value from “0” to “7”. It is a line formed by this.

  Referring to FIG. 16, the relationship between the loop count value and the remaining section maximum calculation time is approximated by a linear function y = 1.1x + 5.9 based on each value shown in FIG. It becomes clear that a value larger than the remaining section maximum calculation time in the variable value can be taken.

  FIG. 17 is a flowchart showing details of the calculation time function approximation processing shown in FIG. 11 by the deriving unit 33 shown in FIG.

  In FIG. 17, the deriving unit 33 executes the processing operations shown in step S122 and the subsequent steps for each of the calculation time measurement positions (calculation time measurement start positions) described above (as shown in FIG. 14). (Step S121). That is, the deriving unit 33 first checks whether the type of calculation amount variation factor related to an arbitrary calculation time measurement position is due to a loop (step S122). As a result of the check, if it is determined that it is due to a loop (YES in step S122), the calculation time is calculated for each value of the number of loops recorded (in the correspondence table 109 shown in FIG. 15) together with the calculation time. The maximum value of the calculation time required from the measurement position to the end of the designated section described above is obtained, and the process of recording in the correspondence table 109 shown in FIG. 15 is executed (step S123).

  Next, based on the correspondence table 109 shown in FIG. 15, the relationship between the individual loop count values and the remaining section maximum calculation time corresponding to each loop count value is, for example, the linear function y shown in FIG. A process of approximating with a linear function such as = 1.1x + 5.9 (so-called function approximation process) is executed, and a process of recording the approximated linear function is executed. Thereby, the process at the calculation time measurement position ends (step S124). As described with reference to FIG. 16, in the function approximation process, approximation is performed for any loop count value using a function that takes a value larger than the maximum calculation time obtained by profile acquisition.

  In general, in the loop described above, the calculation time tends to increase temporarily with respect to the number of loops, and therefore it is predicted that function approximation can be performed using a linear expression (linear function). However, depending on the actually acquired data, approximation in the form of other functions is also possible. When there is a margin in the code size of the program source 39, the correspondence table 109 shown in FIG. 15 is stored in the data table indicating the relationship between the above loop count value and the remaining section maximum calculation time in the program source 39. It is also possible to have as.

  As a result of the check in step S122, if it is determined that the type of the calculation amount variation factor related to the calculation time measurement position is not due to the loop (NO in step S122), the deriving unit 33 obtains the calculation result obtained by the profile acquisition. The maximum value of the calculation time is obtained from all the calculation times from the calculation time measurement position to the end of the specified section. Then, the calculated maximum value of the calculation time is recorded in the correspondence table 109 shown in FIG. 15 as the remaining section maximum time. Thereby, the process at the calculation time measurement position ends (step S125).

  When the processing operations shown in steps S123 to S125 are executed at all the calculation time measurement positions, the series of processing operations shown in FIG. 17 ends.

  FIG. 18 is a flowchart showing an example of processing operations executed in the calculation function adding unit 35 shown in FIG.

  In the flowchart shown in FIG. 18, the calculation function adding unit 35 inserts the necessary CPU resource amount estimation code into the program source 39 input to the program conversion device 9 by the method described below. That is, first, the computer adding unit 35 is the deadline time (that is, the time obtained by adding the execution time given as the performance index to the current time at the start position of the specified section in the program source 39, A process for inserting a code for obtaining the time at which the program source 39 must end is executed (step S131). Next, as a process at each calculation time measurement position described above (step S132), a process for inserting a necessary CPU resource re-estimation code described in detail later is executed at that position. And the process which records the program source produced | generated as a result of inserting the said code is performed (step S133). When the above process is executed for all the calculation time measurement positions, the series of processing operations shown in FIG. 18 is completed.

  FIG. 19 is a flowchart showing an example of processing contents related to a necessary CPU resource amount estimation code executed by the calculation function addition unit 35 shown in FIG.

  In FIG. 19, the calculation function adding unit 35 first checks whether the type of calculation amount variation factor related to an arbitrary calculation time measurement position is due to a loop (step S <b> 141). If it is determined as a result of the check (YES in step S141), the determined loop count value is obtained from the loop count variable “n” in the loop that is the calculation amount variation factor (step S142). Next, the remaining section maximum calculation time is calculated using the loop count value acquired in step S142 and the above approximate function. Here, instead of using the above approximate function, when the remaining section maximum calculation time is obtained using the data table, the remaining section maximum calculation time corresponding to the loop count is calculated from the data table. This is obtained (step S143). When the processing operation is completed, the calculation function adding unit 35 acquires the current time next, and the margin from the acquired current time to the deadline time related to the execution of the processing operation obtained in step S131 of FIG. Processing for obtaining time is executed (step S144).

  When the processing operation is completed, the calculation function adding unit 35 then executes the processing operation in the remaining section by dividing the remaining section maximum calculation time obtained in step S143 by the margin time until the deadline time. CPU resource ratio (that is, necessary CPU resource ratio) necessary to finish the process until the deadline time is obtained (step S145). When the processing operation shown in step S145 ends, the series of processing operations shown in FIG. 19 ends. As a result of the check in step S141, if it is determined that it is not due to a loop (NO in step S141), the process immediately proceeds to the processing operation shown in step S144. In this case, since the remaining section maximum calculation time is known at the time of insertion of the insertion code from the correspondence table 107 shown in FIG. 14, the value recorded in the correspondence table 107 is used.

  FIG. 20 is a flowchart showing an example of the processing operation executed in the assignment function adding unit 37 shown in FIG.

  In FIG. 20, the allocation function adding unit 37 executes the processing operation shown in step S152 for each calculation time measurement position (step S151). That is, when the required CPU resource ratio (calculated by the required CPU resource amount estimation code inserted in the calculation function adding unit 35) obtained for each calculation time measurement position is output from the calculation function adding unit 35, The allocation function adding unit 37 executes a program conversion process so that the necessary CPU resource ratio can be secured (step S152). Then, when the processing operation shown in step S152 is finished for every calculation time measurement position, the series of processing operations shown in FIG. 20 is finished.

  An example of a means for securing the CPU resource ratio used in the processing operation shown in FIG. 20 is the technique disclosed in Non-Patent Document 1 described above. When this method is used, a code for calling a system call for setting the required CPU resource ratio in the resource control mechanism in the OS is inserted.

  FIG. 21 is an explanatory diagram showing an example of a performance ensuring program output from the program conversion apparatus 9 shown in FIG. 2 based on the program source 47 shown in FIG. FIG. 21 shows an example of the performance ensuring program converted from the program source 39 whose file name is “src_b.c”.

In the performance ensuring program 161 shown in FIG. 21, the deadline (DL) time is designated from the current time as indicated by the reference numeral 161a at the position immediately before the line number “3” that is the start position of the designated section described above. A code for setting a time after “20 milliseconds” that is the execution time is inserted. Further, a necessary CPU resource amount estimation code is inserted at a position immediately before the line number “3”, which is a position recorded in the correspondence table 107 shown in FIG. 14, as indicated by reference numeral 161b. The necessary CPU resource ratio (r) at the position indicated by the reference numeral 161b is “14.2 milliseconds” that is the maximum calculation time of the remaining section recorded in the correspondence table 107, and is the time between the current time and the deadline time. It is obtained by dividing by. In addition, a code for ensuring the above (determined) required CPU resource ratio (r) is inserted at a position immediately after the position indicated by the reference numeral 161b, as indicated by a reference numeral 161c. Further, the line number “6” which is the position recorded in the loop list 65 shown in FIG.
A necessary CPU resource amount estimation code is also inserted at a position immediately before, as indicated by reference numeral 161d. The necessary CPU resource ratio (r) at the position indicated by the reference numeral 161d is obtained by substituting the value “n” of the loop count variable into x of the linear function y = 1.1x + 5.9 which is the relationship obtained in FIG. This is obtained by dividing the calculated remaining section maximum calculation time by the time between the current time and the deadline time. In addition, a code for ensuring the above (determined) necessary CPU resource ratio (r) is inserted at a position immediately after the position indicated by the reference numeral 161d, as indicated by a reference numeral 161e.

  The program source 161 having the above-described configuration is output from the program conversion device 9 to the receiver main body 1 as the performance ensuring program of the program source (39) with the file name “src_b.c”.

  The preferred embodiment of the present invention has been described above, but this is an example for explaining the present invention, and the scope of the present invention is not limited to this embodiment. The present invention can be implemented in various other forms.

1 Digital broadcast receiver body (receiver body)
3 Remote controller (remote control)
Reference Signs List 5 antenna 7 display device 9 program conversion device 11 storage device 13 processor 15 signal receiver 17 digital broadcast tuner 19 digital broadcast receiver software 21 remote controller task (remote control task)
23 Tuner Task 25 Video Decoding Task 27 Audio Decoding Task 29 Data Broadcast Browser Task 31 Computation Amount Fluctuation Factor Extraction Unit (Extraction Unit)
33 Derivation unit (derivation unit) for relationship between variable factors and computational complexity
35 Required CPU resource amount calculation function addition part (calculation function addition part)
37 CPU resource allocation function addition unit (allocation function addition unit)
39 Program source 41 Section and execution time request specification code (code)
100 Digital broadcast receiving system

Claims (8)

In a program conversion device for converting a source code of a given program,
Execution time for calculating the maximum time required for execution from one place or a plurality of places in the section to the end of the section for a section of the program to which the performance index of the program is given A calculation unit;
Secured for execution from the one or more places to the end of the section in the CPU where the program starts from the maximum value of the execution time calculated by the execution time calculation unit and the performance index A function adding unit that gives a function for obtaining a CPU resource ratio to be given to the source code of the program;
A program conversion unit that converts the source code of a program function of obtaining the CPU resource ratio by the functionalization unit is granted, the program to ensure the CPU resource percentage determined by the function determining the pre-Symbol C PU resources ratio,
A program conversion apparatus comprising: The program conversion apparatus according to claim 1, wherein
In certain the section of the program performance index of the program is given, further comprising a variation cause detection unit, for detecting whether the occurred caused on the change in the calculated amount of the program,
The program conversion apparatus in which one place or a plurality of places in a certain section of the program is a position where the cause of the change detected by the change cause detection unit exists. The program conversion device according to claim 2, wherein
A program conversion apparatus in which the cause of the variation is a branch destination of a conditional branch existing in the section. The program conversion device according to claim 2, wherein
The program conversion apparatus in which the cause of the variation is a loop existing in the section, and the one place or a plurality of places includes places after the determined number of loops of the loop in the section. The program conversion device according to claim 4, wherein
The program conversion apparatus in which the execution time calculation unit obtains a maximum value of an execution time according to the determined number of loops existing in the section detected by the variation cause detection unit. The program conversion device according to claim 5, wherein
The execution time calculation unit
A generating unit that generates a profile acquisition code for obtaining a time required for execution of the section of the program;
An execution unit for executing the profile acquisition code generated by the generation unit;
A calculation unit that calculates the maximum value of the execution time from profile data acquired by execution of the profile acquisition code by the execution unit;
A program conversion apparatus comprising: The program conversion apparatus according to claim 6, wherein
The program conversion apparatus, wherein the performance index is a time required for executing the section of the program. In a program conversion method for converting a source code of a given program,
A first value for calculating the maximum time required for execution from one place or a plurality of places in the section to the end of the section for a section of the program to which the performance index of the program is given And the steps
Secured for execution from the one or a plurality of places to the end of the section in the CPU that starts the program from the maximum value of the execution time calculated in the first step and the performance index A second step of providing a function for obtaining a CPU resource ratio to be added to the source code of the program;
A third step of converting the program to ensure the source code of the program function is imparted to determine the CPU resource ratio in the second step, CPU resources rate determined by the function determining the pre-Symbol C PU resources ratio When,
A program conversion method comprising:

Images