JP4642398B2 - Shared bus arbitration system - Google Patents

Description

  The present invention relates to a shared bus arbitration system having a shared bus arbitration function shared by a plurality of CPUs (Central Processing Units) and the like.

  As a conventional example of a system including a plurality of CPUs (multi-CPU system), there is a system as shown in FIG. The system in FIG. 8 includes two CPUs 101 and 102, a RAM (Random Access Memory) 103 dedicated to the CPU 101, an input / output port 104 and a dedicated bus 108, and a RAM 105 dedicated to the CPU 102, an input / output port 106 and a dedicated bus 109. The CPU 101 and the CPU 102 are interfaced by a dual port memory 107.

  Normally, the RAM 103 needs to have a memory size when the CPU 101 reaches a maximum load (a state where the largest RAM area is used) so that a memory shortage does not occur in either the CPU 101 or the CPU 102. The RAM 104 needs to have a memory size when the CPU 102 reaches the maximum load.

  If both the CPU 101 and the CPU 102 require input / output ports, it is necessary to provide dedicated input / output ports (input / output ports 104 and 105) for each CPU as shown in FIG. For this reason, the chip size of the entire system increases, leading to an increase in chip cost and a decrease in yield.

  Therefore, a system including an arbitration circuit as shown in FIG. 9 has been put into practical use (for example, see Patent Document 1 below). The CPU 111 and the CPU 112 share the RAM 121 and the input / output port 122 for use. For example, the operation when the CPUs 111 and 112 wish to access the RAM 121 (reading and / or writing data) will be briefly described. First, each of the CPUs 111 and 112 sends access request signals 131 and 132 to the arbitration circuit 117.

  When the access request signals 131 and 132 do not compete, the arbitration circuit 117 gives access permission response signals 133 and 134 to the CPU (CPU 111 or 112) that issued the access request signal to give the right to use the shared bus 120. The CPU 111 or 112 that has acquired the right to use the shared bus 120 accesses the RAM 121 via the dedicated bus 118 or the dedicated bus 119 and the shared bus 120.

  However, when the access request signals 131 and 132 compete, the arbitration circuit 117 gives the right to use the shared bus 120 to the CPU 111 or 112 according to a predetermined priority order. The CPU 111 or 112 that has obtained the use right performs the same operation as when the access request signal does not compete, but the CPU 111 or 112 that has not obtained the use right enters a wait state by the response signal 133 or 134 of the wait request, The execution of the data transfer program is suspended until the usage right can be acquired.

  In the system of FIG. 9 as well, a RAM 113 dedicated to the CPU 111, an input / output port 114 and a dedicated bus 118, and a RAM 115 dedicated to the CPU 112, an input / output port 116 and a dedicated bus 119 are provided.

Japanese Patent Publication No. 06-028052

  The operation example of FIG. 9 is a relatively simple operation example. However, the arbitration circuit 117 is actually only based on the presence / absence of an access request signal from each CPU, the priority for giving the right to use, and the use state of the shared bus 120. In addition, since the contents of the response signals 133 and 134 are determined while referring to the current state of the entire system, it is necessary to generate response signals and control the bus according to a plurality of conditions, and the circuit is complicated. And it has become large-scale.

  As the number of CPUs sharing the shared bus 120 increases, the connection between the dedicated bus and the shared bus increases, and the access request signal receiving circuit and the response signal generating circuit are extremely complex and large-scale. In other words, it is necessary to make a large-scale change to the arbitration circuit 117, and the design time and design cost for changing the system are extremely large.

  Considering that the CPUs 111 and 112 are denied even if they want to access the RAM 121, a RAM 113 having a size close to the memory size required at the maximum load of the CPU 111 is provided exclusively for the CPU 111, and at the maximum load of the CPU 112. A RAM 115 having a size close to a necessary memory size needs to be provided exclusively for the CPU 112. For this reason, even in the system of FIG. 9, the effect of reducing the chip size is not sufficient. Similarly, since the right to use the shared bus 120 is not always obtained at the timing when the input / output port 122 is to be accessed, it is necessary to separately provide the input / output port 114 dedicated to the CPU 111 and the input / output port 116 dedicated to the CPU 112. This also prevents the chip size from being reduced.

  In view of the above points, the present invention realizes arbitration of a shared bus with a small circuit configuration, contributes to a reduction in chip size, and is easy to change the number of CPUs sharing the shared bus. The purpose is to provide a shared bus arbitration system that can handle the above.

  In order to achieve the above object, a shared bus arbitration system according to the present invention includes a system clock generating means for generating a system clock, a plurality of arithmetic means for performing predetermined arithmetic processing, and driving conditions for each of the plurality of arithmetic means. A first register that stores the counter, a shared bus that is shared by the plurality of calculation means and can transmit input / output signals of each of the plurality of calculation means, and a counter that updates a count value in synchronization with the system clock And each computing means obtains the right to use the shared bus when the count value satisfies its own driving condition, and each of the driving conditions has two or more computing means for the same count value. These driving conditions are determined so as not to be satisfied at the same time.

  Thereby, there is no contention for the shared bus. In other words, the shared bus arbitration function is realized by applying a counter function that can be implemented with a relatively small circuit configuration, and there is no need to provide a complex arbitration circuit as in the conventional example.

  In addition, it is possible to change the operating rate of each computing means by modifying the counting method of the count value or changing the stored contents of the first register. Then, when a memory such as a RAM is provided as a peripheral device shared via the shared bus in each calculation means, it is possible to allocate a memory area according to the operation status of each calculation means. Thus, as in the conventional examples shown in FIGS. 8 and 9, the memory size required at the maximum load (or a size close to the memory size required at the maximum load) is calculated for each computing means (corresponding to the CPU in the conventional example). There is no need to provide a dedicated memory. This contributes to a reduction in chip size when the shared bus arbitration system is formed on a semiconductor substrate.

  In addition, when a memory such as a RAM as a peripheral device of each arithmetic means is connected to the shared bus, the memory is shared by the arithmetic means, so that each arithmetic means calculates by accessing the memory. Data transfer between means becomes unnecessary. For this reason, each computing means can maintain the wait state for a longer time than the conventional example, and the power consumption of the system can be reduced.

  Further, if configured as described above, even if the arithmetic means sharing the shared bus changes, the small number of bits such as changing the number of bits of the counter and changing the decoder for decoding the count value. It can be handled with simple changes. That is, it is possible to cope with a change in the number of arithmetic means sharing the shared bus by an easy design change.

  Further, for example, each calculation means may input a pulse when the count value satisfies its own drive condition among pulses constituting the system clock as an operation clock to itself.

  Each calculation means operates by inputting, as an operation clock to itself, a pulse when the count value satisfies its own drive condition among pulses constituting the system clock. Each arithmetic means acquires the right to use the shared bus when the count value satisfies its own drive condition. Therefore, at the timing when the operation clock is input to each arithmetic means, each arithmetic means is always given the right to use the shared bus. Therefore, each arithmetic means uses the shared bus as if it had its own dedicated bus. Can be used.

  That is, when using the shared bus, each computing means does not need to send an access request signal to inquire whether or not the shared bus can be used as in the conventional example, and the use of the shared bus by the operation of other computing means is not required. Whether or not the shared bus can be used is determined only by the count value and its own drive condition. This speeds up and simplifies the processing of the computing means.

  Further, as a specific example, the drive condition is given as a drive number, and the time when the count value satisfies its drive condition corresponds to the time when the count value matches the drive number.

  In addition, for example, a peripheral device that can input data output from each arithmetic means via the shared bus or can output data to each arithmetic means, and a second condition that stores driving conditions of the peripheral device A register, and the peripheral device inputs a pulse when the count value satisfies its drive condition among pulses constituting the system clock as an operation clock to the peripheral device, and the count value is The right to use the shared bus may be acquired when its own drive condition is satisfied.

  For example, if the plurality of calculation means are composed of a first calculation means and a second calculation means, and the driving conditions of the first calculation means and the driving conditions of the peripheral device are matched, the peripheral device is It will operate as a peripheral device used for one computing means. Then, by rewriting the storage contents of the first register to change the driving conditions of the first arithmetic means and the second arithmetic means, or by rewriting the storage contents of the second register, the driving conditions of the peripheral device are changed. By changing the peripheral device, the peripheral device can be operated as a peripheral device used for the second calculation means.

  In other words, each of the driving conditions of the plurality of arithmetic means and / or the driving conditions of the peripheral device may be made variable. That is, each of the driving conditions of the plurality of calculating means is variable, the driving condition of the peripheral device is variable, or each of the driving conditions of the plurality of calculating means and the driving condition of the peripheral device are variable. It is good to do.

  Thereby, the combination of the said peripheral device and each calculating means can be changed freely and easily.

  As described above, according to the automatic time constant adjustment circuit of the present invention, it is possible to realize arbitration of the shared bus with a small circuit configuration and contribute to the reduction of the chip size. Furthermore, it is possible to easily cope with a change in the number of CPUs sharing a shared bus.

<< First Embodiment >>
Hereinafter, a first embodiment of a shared bus arbitration system according to the present invention will be described with reference to the drawings. FIG. 1 is a block diagram of a shared bus arbitration system 10 according to the first embodiment of the present invention.

  The shared bus arbitration system 10 includes CPUs 1, 2, and 3 (arithmetic unit) that perform predetermined arithmetic processing (program execution, data transfer, etc.), respectively, according to tasks performed by the shared bus arbitration system 10, and a semiconductor memory. RAM4 and external data can be input and the data can be output (transmitted) to CPU1, 2 or 3, and the data output by CPU1, 2 or 3 can be input and the data output to the outside An input / output port 5 that can be used, a system clock generation circuit (system clock generation means) 6 that generates a system clock, and a counter 7 that inputs the system clock and updates a count value in synchronization with the system clock; , A signal shared by the CPUs 1, 2 and 3 and sent from the RAM 4 or the input / output port 5 is input to the CPU 1, 2 or 3 From the shared bus 8 for transmitting the output signal from the CPU 1, 2 or 3 to the RAM 4 or the input / output port 5, the arbitration circuits 11, 12 and 13, and the registers 21, 22 and 23. It is roughly structured. The RAM 4 and the input / output port 5 function as peripheral devices of the CPU 1, 2 or 3 shared by the CPUs 1, 2, and 3.

  The system clock generated by the system clock generation circuit 6 is supplied to each of the arbitration circuits 11, 12, and 13, and as the operation clock of each of the RAM 4 and the input / output port 5, is directly applied to each of the RAM 4 and the input / output port 5. Is given. The count value output from the counter 7 is given to each of the arbitration circuits 11, 12 and 13.

  The register 21 stores a drive number for the CPU 1 and gives the drive number to the arbitration circuit 11. Similarly, the registers 22 and 23 store a drive number for the CPU 2 and a drive number for the CPU 3, respectively. The drive numbers are given to the arbitration circuits 12 and 13, respectively. It is assumed that the drive number for CPU1, the drive number for CPU2, and the drive number for CPU3 are 1, 2, and 3, respectively.

  The arbitration circuit 11 extracts a part of the system clock based on the count value supplied from the counter 7 and the drive number for the CPU 1 supplied from the register 21, and uses the part of the system clock as the operation clock 51 for the CPU 1. Supply. Similarly, the arbitration circuit 12 extracts a part of the system clock based on the count value and the drive number for the CPU 2, supplies a part of the system clock to the CPU 2 as the operation clock 52, and the arbitration circuit 13 determines the count value and A part of the system clock is extracted based on the drive number for the CPU 3 and a part of the system clock is supplied to the CPU 3 as the operation clock 53.

  The state of each operation clock will be described with reference to FIG. FIG. 2 shows, from above, the voltage waveform of the system clock, the count value output by the counter 7, the voltage waveform of the operation clock 51 of the CPU 1, the voltage waveform of the operation clock 52 of the CPU 2, and the voltage waveform of the operation clock 53 of the CPU 3. Yes.

  As shown in FIG. 2, the count value is incremented (incremented by 1) in synchronization with the rise (up edge) of the system clock voltage. When the count value is 3, when the system clock voltage rises, Return to zero. That is, the counter 7 is a circular type counter, and the count value is in a loop like 0 → 1 → 2 → 3 → 0 → 1 → 2 → 3 → 0... In synchronization with the system clock. Change.

  Of the pulses constituting the system clock, the pulse when the count value is 1 becomes the operation clock 51 of the CPU 1. That is, of the pulses constituting the system clock, the pulse when the count value matches 1 which is the drive number for the CPU 1 becomes the operation clock 51 of the CPU 1.

  Similarly, of the pulses constituting the system clock, the pulse when the count value matches 2 which is the drive number for CPU 2 is the operation clock 52 of CPU 2, and the count value matches 3 which is the drive number for CPU 3. The pulse at the time of operation becomes the operation clock 53 of the CPU 3.

  In FIG. 1, reference numerals 41, 42, and 43 denote dedicated input / output buses for transmitting signals input / output by the CPUs 1, 2, and 3, respectively. Each input / output terminal of the RAM 4 and the input / output port 5 is connected to the shared bus 8. The CPUs 1, 2, and 3 are respectively connected to the input / output bus 41 and the arbitration circuit 11, and the input / output bus 42 and the arbitration circuit 12. The RAM 4 and the input / output port 5 can be accessed via the input / output bus 43 and the arbitration circuit 13 and further via the shared bus 8.

  The CPU 1 acquires the right to use the shared bus 8 when the count value matches 1 which is the drive number for the CPU 1 by the function of the arbitration circuit 11, while when the count value is different from 1 which is the drive number for the CPU 1. The right to use the shared bus 8 is not held. Here, “acquiring the right to use the shared bus 8” means that “the signal transmitted by the shared bus 8 is transmitted via the arbitration circuit (the arbitration circuit 11 for CPU1) and the input / output bus (the input / output bus 41 for CPU1). Or “can output a desired signal to the shared bus 8 via an input / output bus (input / output bus 41 for CPU1) and an arbitration circuit (arbitration circuit 11 for CPU1)”, It includes that both input and output are possible.

  Similarly, the CPUs 2 and 3 acquire the right to use the shared bus 8 when the count value matches the own drive number by the functions of the arbitration circuits 12 and 13, respectively, while the count value is different from the own drive number. The right to use the shared bus 8 is not held.

  The drive numbers of the CPUs 1, 2, and 3 are different from each other (as described above, the drive numbers of the CPUs 1, 2, and 3 are 1, 2, and 3 respectively), and two or more for the same count value. The drive numbers corresponding to the CPUs are set so as not to match. Therefore, contention for the shared bus 8 does not occur. That is, the arbitration function of the shared bus is realized by a simple circuit configuration including the counter 7 and the arbitration circuit 11 that performs bus connection switching and system clock masking according to the count value and the stored contents of the register 21 and the like.

  Further, if the counting method of the count value synchronized with the rise of the voltage waveform of the system clock is modified, the operating rate of each CPU can be easily changed. For example, when CPU 1 has the maximum load (a state in which the largest RAM area is used) and CPU 2 and CPU 3 do not require so much RAM area, the CPU 1 counts in synchronization with the rise of the voltage waveform of the system clock. The count value counting method is modified so that the value changes as 0 → 1 → 1 → 1 → 1 → 2 → 3 → 0 → 1 → 1 → 1 → 1 → 2 → 3 → 0. As a result, the CPU 1 can use the RAM 4 preferentially over the CPUs 2 and 3, and each CPU can secure a required RAM area. In this case, of course, the CPU 1 can also use the input / output port 5 with priority over the CPU 2 and CPU 3.

  Further, in synchronization with the rise of the voltage waveform of the system clock, the count value changes in a loop from 0 to 6 as 0 → 1 → 2 → 3 → 4 → 5 → 6 → 0 → 1 → 2 →. The CPU 1 drive number stored in the register 21 is set to 1 and 4, the CPU 2 drive number stored in the register 22 is set to 2 and 5, and the CPU 3 drive number stored in the register 23 is set. May be 3 and 6. When the CPU 1 has a maximum load and the CPU 2 and the CPU 3 do not require so much RAM area, the storage contents of the registers 21, 22 and 23 are changed by a drive number changing circuit (not shown) to Are set to 1, 2, 3, and 4, the drive number for CPU 2 is set to 5, and the drive number for CPU 3 is set to 6. This also makes it possible for the CPU 1 to use the RAM 4 preferentially over the CPUs 2 and 3, and each CPU can secure the required RAM area. In this case, of course, the CPU 1 can also use the input / output port 5 with priority over the CPU 2 and CPU 3.

  As described above, the operating rate of each CPU can be easily increased by modifying the counting method of the count value synchronized with the rise of the voltage waveform of the system clock or by changing the stored contents of the registers 21, 22 and 23. The RAM area can be distributed according to the operating status of each CPU.

  When a system such as the system shown in FIGS. 8 and 9 (conventional example) or the shared bus arbitration system 10 is formed on a semiconductor substrate, the size of the RAM greatly affects the chip size. According to the arbitration system 10, since the RAM area can be allocated according to the operation status of each CPU, the memory size required at the maximum load (or the memory size required at the maximum load) for each CPU as in the conventional example. There is no need to provide a dedicated RAM having a size close to Therefore, the semiconductor chip size can be reduced. Further, the CPUs 1, 2, and 3 also share the input / output port 5, which contributes to a reduction in chip size.

  Further, in the system of FIG. 9 (conventional example), data not stored in the RAM 121 needs to be transferred to the RAM 121 using a transfer function such as a DMA (direct memory access) function (a CPU using the RAM 121). Need to transfer data between). However, in the shared bus arbitration system 10, since the RAM 4 is shared by the CPUs 1, 2, and 3, data transfer becomes unnecessary when each CPU accesses the RAM 4. Therefore, each CPU can maintain the wait state for a longer time than the conventional example, thereby reducing the power consumption of the system.

  Each CPU operates by inputting, as its own operation clock, a pulse “when the count value matches its own drive number” among pulses constituting the system clock. Each CPU obtains the right to use the shared bus 8 when “the count value matches its own drive number”. Accordingly, at the timing when the operation clock is input to each CPU, the right to use the shared bus 8 is always given to each CPU, so that each CPU makes the shared bus 8 as if it were its own dedicated bus. Can be used.

  That is, when using the shared bus 8, each CPU does not need to send an access request signal to inquire whether or not the shared bus 8 can be used as in the conventional example, and the operations of the CPUs 2 and 3 which are other CPUs do not need to be inquired. Whether or not the shared bus 8 can be used does not fluctuate (whether or not the shared bus 8 can be used is determined only by the count value and its own drive number). As a result, the processing speed of the CPU can be increased and simplified.

  When the count value is 0, 2 or 3, and the count value does not match the drive number for the CPU 1, the arbitration circuit 11 stops outputting the voltage to the shared bus 8. Similarly, when the count value does not match the drive number for the CPU 2, the arbitration circuit 12 stops outputting the voltage to the shared bus 8, and when the count value does not match the drive number for the CPU 3, the arbitration circuit 13 shares The voltage output to the bus 8 is stopped.

  This is because the arbitration circuits 11, 12, and 13 are each provided with a three-state buffer that is controlled by the match / mismatch of the drive number for CPU1, the drive number for CPU2, and the drive number for CPU3 with the count value. As a result, the output signals from the arbitration circuits 11, 12, and 13 do not collide with each other.

  In order to prevent the output signals from colliding with each other, the shared bus arbitration system 10 in FIG. 1 may be modified like the shared bus arbitration system 10a in FIG. 3, the same parts as those in FIG. 1 are denoted by the same reference numerals, and detailed description thereof is omitted. The arbitration circuit 60 includes a multiplexer 61 connected to each of the input / output buses 41, 42, and 43. The multiplexer 61 connects the input / output bus 41 to the shared bus 8 when the count value is 1, and the count value is 2 At this time, the input / output bus 42 is connected to the shared bus 8, and when the count value is 3, the input / output bus 43 is connected to the shared bus 8. Thereby, collision of output signals is prevented without using a three-state buffer.

  The arbitration circuit 60 inputs the system clock and count value, and the drive numbers for the CPUs 1, 2, and 3 stored in the registers 21, 22, and 23, respectively, and similarly to the arbitration circuits 11, 12, and 13 in FIG. , Operating clocks 51, 52, and 53 are given to the CPUs 1, 2, and 3, respectively.

<< Second Embodiment >>
Next, a second embodiment of the shared bus arbitration system according to the present invention will be described with reference to the drawings. FIG. 4 is a block diagram of a shared bus arbitration system 10b according to the second embodiment of the present invention. 4, the same parts as those in FIG. 1 are denoted by the same reference numerals, and detailed description thereof is omitted. 4, the relationship among the CPUs 1, 2, 3, the system clock generation circuit 6, the counter 7, the shared bus 8, the arbitration circuits 11, 12, 13 and the registers 21, 22, 23 are as shown in FIG. It is the same as that.

  In the shared bus arbitration system 10b in FIG. 4, arbitration circuits 14 and 15 and registers 24 and 25 are newly provided. The input / output signals of the RAM 4 are transmitted to the common bus 8 via the dedicated input / output bus 44 and the arbitration circuit 14, and the input / output signals of the input / output port 5 are transferred to the shared bus via the dedicated input / output bus 45 and the arbitration circuit 15. 8 is transmitted.

  The system clock generated by the system clock generation circuit 6 is supplied to each of the arbitration circuits 14 and 15. The count value output from the counter 7 is also given to each of the arbitration circuits 14 and 15.

  The register 24 stores a drive number for the RAM 4 and gives the drive number to the arbitration circuit 14. Similarly, the register 25 stores a drive number for the input / output port 5 and gives the drive number to the arbitration circuit 15. Now, the drive number for CPU1, the drive number for CPU2, and the drive number for CPU3 are 1, 2, and 3 as in the first embodiment, and the drive number for RAM4 and the drive for input / output port 5 are the same. Assume that the numbers are 1 and 2, respectively.

  Similar to the arbitration circuits 11, 12 and 13, the arbitration circuit 14 extracts a part of the system clock based on the count value supplied from the counter 7 and the drive number for the RAM 4 supplied from the register 24. A part of the clock is supplied to the RAM 4 as the operation clock 54. Similarly, the arbitration circuit 15 extracts a part of the system clock based on the count value and the drive number for the input / output port 5 and supplies a part of the system clock to the input / output port 5 as the operation clock 55.

  The state of each operation clock will be described with reference to FIG. 5 shows, from above, the voltage waveform of the system clock, the count value output by the counter 7, the voltage waveform of the operation clock 51 of the CPU 1, the voltage waveform of the operation clock 52 of the CPU 2, the voltage waveform of the operation clock 53 of the CPU 3, and the RAM 4 The voltage waveform of the operation clock 54 and the voltage waveform of the operation clock 55 of the input / output port 5 are shown.

  The voltage waveforms of the operation clocks 51, 52, and 53 in relation to the system clock and the count value are the same as those in FIG. Since the drive number for CPU 1 and the drive number for RAM 4 are the same, and the drive number for CPU 2 and the drive number for input / output port 5 are the same, as shown in FIG. The voltage waveform and the voltage waveform of the operation clock 54 of the RAM 4 are the same, and the voltage waveform of the operation clock 52 of the CPU 2 and the voltage waveform of the operation clock 55 of the input / output port 5 are the same.

  The RAM 4 acquires the right to use the shared bus 8 when the count value matches 1 which is the drive number for the RAM 4 by the function of the arbitration circuit 14, while when the count value is different from 1 which is the drive number for the RAM 4. There is no right to use the shared bus 8. Similarly, the input / output port 5 acquires the right to use the shared bus 8 when the count value matches its own drive number by the function of the arbitration circuit 15, while the shared bus 8 when the count value is different from its own drive number. Eight use rights are not held.

(FIG. 6, FIG. 7: Example of writing operation to RAM 4)
In the shared bus arbitration system 10b configured as shown in FIG. 4, the CPU 1 sets the addresses of the RAM 4 to 0x10H and 0x11H (hexadecimal display, 16 and 17 respectively in decimal display), and 8-bit data 0x0AH and 0x0BH (16 The operation of writing decimal numbers and writing 10, 11) in decimal numbers will be described with reference to FIGS.

  FIG. 6 is a block diagram of the shared bus arbitration system 10b showing in detail the CPU1 and RAM4 portions of the CPU1, CPU2, CPU3, RAM4, and input / output port 5 constituting the shared bus arbitration system 10b of FIG. is there. In FIG. 6, the same parts as those in FIG. 4 are denoted by the same reference numerals. In FIG. 6, the description of the CPU 2, the CPU 3 and the input / output port 5, the arbitration circuits 12, 13, and 15, and the registers 22, 23, and 25 are omitted in order to prevent the illustration from becoming complicated.

  The input / output bus 41 dedicated to the CPU 1 includes a dedicated address bus 41a dedicated to the CPU 1, a dedicated data bus 41b, and a dedicated write signal line 41c. The RAM4 dedicated input / output bus 44 includes a dedicated address bus 44a dedicated to the RAM4, a dedicated data bus 44b, and a dedicated write signal line 44c. The shared bus 8 includes a shared address bus 8a, a shared data bus 8b, and a shared write signal line 8c.

  The dedicated address bus 41a, the dedicated data bus 41b, and the dedicated write signal line 41c dedicated to the CPU 1 are connected to the shared address bus 8a, the shared data bus 8b, and the shared write signal line 8c through the arbitration circuit 11, respectively. The dedicated address bus 44a, the dedicated data bus 44b, and the dedicated write signal line 44c are connected to the shared address bus 8a, the shared data bus 8b, and the shared write signal line 8c through the arbitration circuit 14, respectively.

  In FIG. 7, the horizontal axis represents time, from the top, the voltage waveform of the system clock, the count value, the voltage waveform of the operation clock 51 of the CPU 1, the state of the dedicated address bus 41a dedicated to the CPU 1 (signal transmission state), and the dedicated data bus 41b. State, the state of the dedicated write signal line 41c, the state of the shared address bus 8a, the state of the shared data bus 8b, the state of the shared write signal line 8c, the voltage waveform of the operation clock 54 of the RAM 4, the state of the dedicated address bus 44a dedicated to the RAM 4 The state, the state of the dedicated data bus 44b, the state of the dedicated write signal line 44c, the stored content of the RAM 4 at the address 0x010H, and the stored content of the RAM 4 at the address 0x011H are shown.

  7, the count value, the voltage waveform of the operation clock 51 of the CPU 1 and the voltage waveform of the operation clock 54 of the RAM 4 in the relationship with the voltage waveform of the system clock are the same as those in FIG.

  At the timing t1, when the voltage waveform of the system clock rises and the count value is incremented from 0 to 1, the CPU 1 operates with the operation clock 51 input, and writes the data 0x0AH to the address 0x10H of the RAM 4 An address signal of 0x10H is sent to the bus 41a and a data signal of 0x0AH is sent to the dedicated data bus 41b, and the signal to the dedicated write signal line 41c is switched from high level (high potential) to low level (low potential).

  These 0x10H address signals sent to the dedicated address bus 41a, 0x0AH data signals sent to the dedicated data bus 41b, and low level signals sent by switching to the dedicated write signal line 41c are counted from the timing t1. Until the next time t2 when the signal becomes 2 is transmitted to the shared address bus 8a, the shared data bus 8b, and the shared write signal line 8c as they are, and the shared address bus 8a, the shared data bus 8b, and the shared write signal line 8c, respectively. , And further via the arbitration circuit 14, they are transmitted as they are to the dedicated address bus 44 a, the dedicated data bus 44 b and the dedicated write signal line 44 c of the RAM 4.

  On the other hand, between the timing t2 and t3 when the count value changes from 2 → 3 → 0, the address signal of 0x10H sent to the dedicated address bus 41a, the data signal of 0x0AH sent to the dedicated data bus 41b, and the dedicated write Low level signals sent to the signal line 41c are not transmitted to the shared address bus 8a, shared data bus 8b, and shared write signal line 8c, respectively.

  Similarly, at time t3, when the voltage waveform of the system clock rises and the count value is incremented from 0 to 1, the CPU 1 operates with the operation clock 51 input, and writes data 0x0BH to the address 0x11H of the RAM 4 The 0x11H address signal is sent to the dedicated address bus 41a, the 0x0BH data signal is sent to the dedicated data bus 41b, and the signal to the dedicated write signal line 41c is maintained at a low level.

  These 0x11H address signals sent to the dedicated address bus 41a, 0x0BH data signals sent to the dedicated data bus 41b, and low level signals sent to the dedicated write signal line 41c are counted next time from timing t3. Until the timing t4 at which the signal becomes 2 is transmitted to the shared address bus 8a, the shared data bus 8b, and the shared write signal line 8c, respectively, and via the shared address bus 8a, the shared data bus 8b, and the shared write signal line 8c, respectively. Then, they are further transmitted to the dedicated address bus 44a, the dedicated data bus 44b, and the dedicated write signal line 44c of the RAM 4 through the arbitration circuit 14 as they are.

  On the other hand, until the count value becomes 1 next time after timing t4, the 0x11H address signal sent to the address bus 41a, the 0x0BH data signal sent to the dedicated data bus 41b, and the dedicated write signal line 41c. The transmitted low level signals are not transmitted to the shared address bus 8a, shared data bus 8b, and shared write signal line 8c, respectively.

  At the falling timing of the operation clock 54 of the RAM 4 between timings t1 and t2, the RAM 4 receives the address signal 0x10H transmitted to the dedicated address bus 44a of the RAM 4 and the data signal 0x0AH transmitted to the dedicated data bus 44b. In response to the low level signal transmitted to the dedicated write signal line 44c, the data 0x0AH is written to its own address 0x10H.

  Similarly, at the falling timing of the operation clock 54 of the RAM 4 between timings t3 and t4, the RAM 4 transmits the address signal 0x11H transmitted to the dedicated address bus 44a of the RAM 4 and the data signal transmitted to the dedicated data bus 44b. In response to 0x0BH and the low-level signal transmitted to the dedicated write signal line 44c, the data 0x0BH is written to its own address 0x11H.

  As described above, when the count value coincides with 1 that is its drive number (between timings t1 and t2 and between t3 and t4), the CPU 1 shares the shared address bus 8a, the shared data bus 8b, and the shared write signal. The right to use the shared bus 8 including the line 8c is acquired, and among the pulses of the system clock, the pulse at the time when the count value coincides with its drive number 1 is input as the operation clock 51 to itself, It operates in synchronization with the operation clock 51.

  The RAM 4 also acquires the right to use the shared bus 8 including the shared address bus 8a, the shared data bus 8b, and the shared write signal line 8c when the count value coincides with its drive number 1, and the system clock. Among these pulses, a pulse when the count value coincides with its own drive number 1 is input as an operation clock 54 to itself, and operates in synchronization with the operation clock 54. Thereby, the RAM 4 functions as a RAM provided for the CPU 1. The CPU 2 and the CPU 3 cannot use the RAM 4 as their own RAM.

  Similarly, since the drive number for CPU 2 and the drive number for input / output port 5 are the same “2”, CPU 2 and input / output port 5 each acquire the right to use shared bus 8 when the count value is 2. At the same time, the operation is performed by inputting operation clocks 52 and 55 whose voltage waveforms rise and fall at the same timing. Thereby, the input / output port 5 functions as an input / output port provided for the CPU 2. The CPU 1 and CPU 3 cannot use the input / output port 5 as their own input / output port.

  For example, when the RAM 4 is desired to function as the RAM of the CPU 3, the registers 21 and 23 are controlled so that the drive number for the CPU 1 and the drive number for the CPU 3 become 3 and 1 respectively by a drive number changing circuit (not shown). It is only necessary to rewrite the stored contents or rewrite the stored contents of the register 24 so that the drive number for the RAM 4 is 3. For example, the same applies when the input / output port 5 is desired to function as the input / output port of the CPU 1 or CPU 3. That is, the combination of the peripheral device of the RAM 4 or the input / output port 5 and the CPU 1, 2, or 3 can be freely and easily changed.

  Further, after the timing t4 in FIG. 7, the stored contents of the registers 21 and 23 are rewritten so that the drive number for the CPU 1 and the drive number for the CPU 3 are 3 and 1, respectively, or the drive number for the RAM 4 is 3 If the stored contents of the register 24 are rewritten so that the CPU 3 reads the stored contents of the addresses 0x10H and 0x11H of the RAM 4 after that, transfer of two data (0x0AH and 0x0BH) from the CPU 1 to the CPU 3 is realized. Will be. That is, data transfer between CPUs is realized by changing the CPU drive number or the RAM drive number.

(Drive number and drive conditions)
Further, the register 21 in the first embodiment and the second embodiment may be modified to store the driving condition for the CPU 1 instead of the driving number for the CPU 1. For example, the driving condition for the CPU 1 indicating that “the count value is 1 or less” is stored in the register 21. This driving condition is given to the arbitration circuit 11. The arbitration circuit 11 causes the CPU 1 to acquire the right to use the shared bus 8 when the count value satisfies the drive condition for the CPU 1, that is, when the count value is 0 or 1, while the count value is the drive condition for the CPU 1. When the above condition is not satisfied, the right to use the shared bus 8 is not given to the CPU 1.

  This modification means that “driving number” in the first and second embodiments described above is read as “driving condition”, and the content that means “the count value matches the driving number” is “counted”. This is equivalent to rereading with the meaning of “value satisfies drive condition”.

  Similarly, the registers 22, 23, 24, and 25 may be modified to store the driving conditions for the CPU 2, the driving conditions for the CPU 3, the driving conditions for the RAM 4, and the driving conditions for the input / output port 5, respectively. Good. In this case as well, the CPUs 1 and 2 do not conflict with the shared bus 8, that is, the drive conditions corresponding to two or more CPUs are not simultaneously satisfied for the same count value. Each of the three driving conditions is defined. For example, the driving condition for CPU1, the driving condition for CPU2, and the driving condition for CPU3 are defined as “count value equal to or less than 1,” “count value matches 2”, and “count value matches 3”, respectively.

  Further, for example, when the count value is only a numerical value from 0 to 3 as shown in FIGS. 2 and 5, the driving condition for CPU 1, the driving condition for CPU 2, and the driving condition for CPU 3 are respectively set to “count value”. It is also possible to prevent the CPU 3 from operating at all by defining “1 or less”, “count value is 2 or 3”, and “count value matches 255”.

<< Other, deformation, etc. >>
Further, the present invention is not limited to the case where there are three CPUs as in the above-described embodiment, and can be widely applied to a system in which a shared bus is shared by two or more CPUs. Even if the number of shared CPUs changes (especially increases), it can be handled with a small change that changes the number of bits of the counter 7 and changes the decoder (included in the arbitration circuit 11 etc.) that decodes the count value. Is possible. In other words, it is possible to cope with a change in the number of CPUs sharing the shared bus by an easy design change.

  The CPUs indicated by reference numerals 1, 2, and 3 may be replaced by MPUs (Micro Processing Units), DSPs (Digital Signal Processors), and the like, respectively, depending on the tasks performed by the shared bus arbitration system 10 and the like. Any arithmetic circuit that performs predetermined arithmetic processing (execution of a program, data transfer, etc.) may be used.

  In the above embodiment, the RAM 4 and the input / output port 5 are illustrated as peripheral devices of the CPU 1, 2, or 3 shared by the CPUs 1, 2, and 3. However, the peripheral device shared by the CPUs 1, 2, and 3 is not limited to the RAM and the input / output port, and the data output by the CPUs 1, 2, or 3 can be input via the shared bus 8, and Any peripheral device capable of outputting data to the CPU 1, 2, or 3 is acceptable. For example, a ROM (Read Only Memory) may be used, and a DSP may be a peripheral device.

  The count value may be modified so as to change (increment) in synchronization with the falling edge of the system clock.

  The present invention is suitable for semiconductor devices such as SOC (System On a Chip). In addition, the present invention is suitable for various electric devices such as mobile communication devices such as mobile phones and PHS (Personal Handyphone System) that require low power consumption, and information processing devices represented by personal computers.

1 is a block configuration diagram of a shared bus arbitration system according to a first embodiment of the present invention. It is a figure for demonstrating the relationship between the counter output in FIG. 1, and the voltage waveform of each part. It is a block block diagram which shows the modification of FIG. It is a block block diagram of the shared bus arbitration system which concerns on 2nd Embodiment of this invention. FIG. 5 is a diagram for explaining a relationship between a counter output in FIG. 4 and a voltage waveform of each part. It is the block block diagram which showed a part of FIG. 4 in detail. It is a figure for demonstrating operation | movement of the shared bus arbitration system of FIG. It is a figure which shows the structure of the conventional multi CPU system. It is a figure which shows the structure of the other conventional multi CPU system.

Explanation of symbols

1, 2, 3 CPU
4 RAM
DESCRIPTION OF SYMBOLS 5 Input / output port 6 Clock generation circuit 7 Counter 8 Shared bus 10, 10a, 10b Shared bus arbitration system 11, 12, 13, 14, 15, 60 Arbitration circuit 21, 22, 23, 24, 25 Register 41, 42, 43 44, 44 I / O bus 51, 52, 53, 54, 55 Operation clock 61 Multiplexer

Claims (4)

System clock generating means for generating a system clock;
A plurality of calculation means for performing predetermined calculation processing;
A first register for storing driving conditions of each of the plurality of computing means;
A shared bus shared by the plurality of computing means and capable of transmitting respective input / output signals of the plurality of computing means;
A counter that updates a count value in synchronization with the system clock;
Peripheral devices that can input data output by each computing means via the shared bus, or that can output data to each computing means,
A second register for storing driving conditions of the peripheral device ,
The count value with commonly applied to the peripheral device and the computing means,
Each of the arithmetic means and the peripheral device inputs, as an operation clock to itself, a pulse when the count value satisfies its drive condition among pulses constituting the system clock, and the count value Does not operate when it does not satisfy its own driving conditions, and when the count value satisfies its own driving conditions , operates itself based on the system clock and acquires the right to use the shared bus,
Each of the driving conditions of the plurality of calculating means is determined such that the driving conditions of two or more calculating means are not simultaneously satisfied for the same count value ,
The shared bus arbitration system is characterized in that the drive condition of the peripheral device and one drive condition of the calculation means are determined to satisfy the same count value at the same time .   The drive condition is given as a drive number, and the time when the count value satisfies its drive condition corresponds to the time when the count value matches the drive number. Shared bus arbitration system.   2. The shared bus arbitration system according to claim 1, wherein each of the driving conditions of the plurality of arithmetic units and / or the driving conditions of the peripheral device is variable.   An electric apparatus comprising the shared bus arbitration system according to any one of claims 1 to 3.

Images