JPS62221040A - Stack control system - Google Patents

Abstract

PURPOSE:To evade a stack fault and to increases the processing speed of a stack control system by detecting previously the stack of a process to be newly set under a run state, and loading this stack as long as it is not included in a real stack. CONSTITUTION:A scheduling table 10 is provided to detect the process to be executed next and a desired block. The table 10 contains the name of the process under execution, the name of the process to be executed next, the block numbers of these processes, and the name and block number of the process to be executed after the next one written in a so-called matrix form. Thus the stack number of the process to be executed next is obtained and the presence or absence of the relevant number is detected by a stack control table 11. If said number is not detected, an undesired block is detected out of a real stack 12 and the stack of the process to be needed in the near future is transcribed to the stack 12 from a main memory 2. As a result, the next data processing can be executed with high efficiency.

Description

[Detailed description of the invention] [Table of contents] Overview Industrial field of application Prior art (Figures 5 to 7) Problems to be solved by the invention Means for solving the problems (Figure 1) Effect Embodiment (a) Structure of the embodiment (Figs. 2 and 3) (b) Operation of the embodiment (Fig. 4) Effects of the invention [Summary] In a virtual stack system, a new run (R
By detecting the stack of a process that should be in the UN) state in advance and loading it when the stack does not exist in the real stack, stack faults can be avoided and processing speeded up. be.

[Industrial application field]

The present invention relates to a stack control method, and particularly to a system using a virtual stack method, in which the number of occurrences of stack faults is reduced and overall processing speed is increased.

[Conventional technology]

When performing data processing using the virtual stack method, for example, the fifth
As shown in the figure, processes Pl, P2, --- Pm ---
Each process has N as a stack area in its own virtual space.
It has a block area. These stack areas are stored in the main memory as shown in FIG.

By the way, these processes P ISP 2---P
When m is operated by a central processing unit (CPU), as shown in Figure 6, the stack area of the required process must exist on the real stack (physical stack) provided within the CPU. It is.

However, this real stack provided on the CPU is n (
There are only blocks (n<N (in the example, n=4 in the figure)), and therefore each process sequentially uses them.

Therefore, at a certain point, physical blocks O to 3 of each real stack
In order to manage how the CPU is used, a stack management table is provided on the CPU side, as shown in FIG. The process name, its block name, valid bin tv, etc. are entered in this stack management table.

When a certain process is in a running state, if a required logical block is not on this real stack, the CPU is notified as an interrupt, and as shown in Figure 7, the data of an appropriate block, for example, physical block 2, is transferred to the main memory. Stack fault processing is performed, in which the data is flushed to the stack and necessary data is read from the stack area of main memory.

Typically, the data transfer associated with this stuck fault is
This is a significant overhead for U.

[Problem that the invention seeks to solve]

Previously, the required block was simply read when a stack fault was detected, that is, the stack block required by the process to be executed did not exist on the real stack. In this case, there are many blocks of processes that are currently running and processes that have recently been running on the actual stack, and when a process switch occurs, there are few stacks of processes that will newly become running. Therefore, there was a problem in that a stack fault occurred every time a process switch was performed.

SUMMARY OF THE INVENTION In order to solve this problem, it is an object of the present invention to provide a stack control method in which stack blocks required by a process are transferred onto the real stack before the process is scheduled. .

[Means for solving problems]

To achieve the above object, in the present invention, as shown in FIG. 1, a scheduling table is provided to detect the process to be executed next and the blocks it requires. In the scheduling table, the name of the process currently being executed, the name of the process to be executed next and its stack block number, the name of the process to be executed next and the block number, etc. are entered in a so-called queue-like format. There is.

[Effect]

Therefore, since the stack number of the process to be executed next is known from the scheduling table, its presence or absence is detected from the stack management table, and if it is not present, unnecessary blocks are searched from the actual stack to create the stack of the process that will be needed in the near future. Since the data is transferred from the main memory to the real stack, the next data processing can be executed efficiently.

〔Example〕

(al) Configuration of an Embodiment The configuration of an embodiment of the present invention will be explained with reference to FIGS. 2 and 3.

FIG. 2 is a configuration diagram of an embodiment of the present invention, and FIG. 3 is an explanatory diagram of its scheduling table.

In FIG. 2, 1 is a CPU, 2 is a main memory, and 3 is a control section.

The CPU 1 has a scheduling table 10, a stack management table 11, a real stack 12, and a transfer flag 1.
3, etc., and the main memory 2 is provided with stack bank up areas 21, 22, 23, 24, 25, etc., and the stack areas of processes P1, P2, P3, and so on are blocks. Stored in units.

The control unit 3 refers to the scheduling table 10 and the stack management table 11 and constantly monitors whether or not there is a block on the real stack that will be used by the process that should become the next run. is used to transfer data between the main memory and the real stack, and at the same time update the stack management table as appropriate. Also, for consistency with the CPUI, the transfer flag 13 is turned on and off, and remains on while the control unit 3 is performing this transfer.
While on, the CPUI does not operate the stack management table 11 to avoid conflict with a stack fault occurring on the CPLI side.

The scheduling table IO should be a running process at one point and then thin (READY state).
It stores information about a process and the stack blocks used when the process is in a run, and is managed by a scheduler (not shown).

The stack management table 11 is a table that holds data such as which block of which process each block O to 3 of the real stack 12 corresponds to, or whether it is not used by any process, and is updated by the CPUI. managed.

(b) Operation of the Embodiment The operation of the embodiment of the present invention shown in FIG. 2 will be explained based on the flowchart showing the operation of the control section 3 shown in FIG. 4 and with reference to a map. The operation of the control unit 3 is shown in FIG. 4(a).
As shown in (), this is a loop that constantly monitors the usage status of the stack and transfers data to the real stack 12 as necessary.

(2) The control unit 3 reads the scheduling table 10 and recognizes the block of the process that will be the next run.

■ The block recognized in this way is the real stack 12
The control unit 3 searches the stack management table 11 to find out whether the stack exists on the top. As a result, if the block already exists on the real stack 12, the control unit 3
performs the above step (2) again, but if the block does not exist, the stack management table 11 is searched for a block of the real stack necessary to load it into the real stack 12 in advance.

■ That is, as shown in FIG. 4(b), first the valid flag ■
Check to see if there are free blocks, and if not, check to see if there are any blocks for non-running processes.
If so, it is checked whether there is a block of a so-called non-ready process in which the block of the non-run process will not become the next run.

■ As a result, if there is no block on the real stack 12 that can transfer the stack necessary for the next process, then
Return to By the way, such a block is the real stack 12
If it is above, the control unit 3 turns on the transfer flag 13.

■ At this time, if the transferable block on the real stack 12 is a free block and no eviction processing is necessary, the control unit 3 reads the block of the next thin process recognized in the above ■ from the main memory 2. is read out and transferred to the real stack 12, and the transfer flag is turned off. However, if valid data exists in the transferable block,
After performing the process of expelling this to main memory 2,
Transfer the block to the real stack and turn off the transfer flag.

〔Effect of the invention〕

According to the present invention, it is possible to detect that the stack necessary for the process is not on the real stack before the process starts running, and to perform the transfer process.

Therefore, when a process enters a running state, it can be expected that the necessary stack will always be on the real stack. Therefore, the occurrence of stack faults can be significantly suppressed, and data processing in the entire system can be performed at high speed.

[Brief explanation of drawings]

FIG. 1 is an explanatory diagram of the principle of the present invention; FIG. 2 is a configuration diagram of an embodiment of the present invention; FIG. 3 is an explanatory diagram of a scheduling table; FIG. 4 is an explanatory diagram of the operation of the control unit in the present invention; FIG. 5 is an explanatory diagram of the virtual space and stack area of a process, FIG. 6 is an explanatory diagram of the virtual stack and real stack, and FIG. 7 is an explanatory diagram of stack fault processing. ■--Central processing unit 2--Raw memory 3-Control unit

Claims (1)

[Scope of Claims] A virtual stack type information processing system that has a real stack independent of main memory, divides this real stack into a plurality of blocks, and transfers data to and from the main memory in units of blocks, A control unit (3) that controls data transfer between main memory and the real stack independently of the central processing unit, a management table (11) that manages multiple blocks of the real stack, and a scheduler that manages the run status of processes. - Equipped with a table (10), the control unit (3) uses the scheduling table (10) to recognize the stack block required for the next run process, and when this block does not exist on the real stack, A stack control method characterized by transferring this onto the real stack.