JPH0754497B2 - Shared I / O device control method - Google Patents

Description

The present invention relates to a control method for a shared input / output device, and more specifically, it outputs only an input / output request that has the highest priority individually and at a time point. When multiple I / O requests from each client compete in a system in which multiple independent devices (hereinafter referred to as clients) that do not output new I / O requests until the processing is complete share the I / O devices. The present invention relates to a control method for controlling the priority order of these input / output requests.

(Prior Art) Conventionally, in a shared system of this type of input / output device, when a plurality of I / O requests from each client compete within a short time, the input / output is performed according to the priority of these requests. A priority control method for controlling requests is adopted.

That is, in FIG. 3, 100 is a shared input / output device, 20
0,300,400 is a client such as a microprocessor that individually outputs a single I / O request to the shared I / O device 100 at a time and does not output a new I / O request until the processing corresponding to the request is completed. is there. These clients 200, 300, 400, when making an input / output request to the shared input / output device 100, add their priorities 21, 31, 41 to the current requests 201, 301, 401, respectively, and input / output requests 210, 310, 410.
Is being output.

Then, the scheduler 101 in the shared input / output device 100 schedules the input / output requests 210, 310, 410 based on the priorities 21, 31, 41, and the priority 21 is the highest, and the order of 31, 41 is the following. Request 201 → 301 → 4 if
The input / output processing 102 for the input / output medium 500 is executed according to a schedule in the order of 01.

(Problems to be solved by the invention) However, according to this control method, each client 200, 30
Requests 201 and 2 that occur sequentially at time 0 and 400 respectively
If the priority of 02, ..., 301,302, ..., 401,402, ... is, for example, requests 201>202>301> ... (the left one has higher priority), the highest priority request 20
If there is a time lag between the client 200 knowing that the process for 1 has been completed by some means before outputting the next request 202, another client will be in between.
Processing for request 301 in 300 was preceded,
Order different from priority, ie request 201 → 301 → 202
→ There is a problem that processing is performed in order.

For example, as shown in FIG. 5, at the timing of the completion of the input / output processing corresponding to one input / output request in the shared input / output device 100, the current request 20
Assuming that the I / O processing corresponding to 1 is completed, the shared I / O device 100 at this point sends the next request 20 to the client 200.
2 does not know that what is happening is that the next request 302 of another client 300 as shown in d is the current request 301.
Will start the corresponding processing.

In the figure, 2A (10), 3A written above the arrow indicating the current request 201, 301, 401 and the next request 202, 302
(1), 4A (3), 2B (9), 3B (5), and the meaning of 10, 9, 1, 5, 3 written above the arrow indicating priority 21, 31, 41 However, it is as in the legend shown in the margin of the figure. For example, 2A (10) is the first request in the client 200 and has the priority value of 10. Similarly, 3B (5) refers to the second request in the client 300 with a priority value of 5, and 4A (3)
Indicates the first request in the client 400 and the priority value is 3. Here, the larger the priority value, the higher the priority.

As a result of the above, as shown in the bottom of FIG. 5, the input / output processing time is 3B (5) → 2B between the timings and.
The processing is executed in the order of (9). In other words, the request 202 of the client 200 that should be processed first in terms of priority
The request 302 (3B of the client 300 rather than (2B (9))
(5)) will be processed first.

In such a case, if it is attempted to perform control for satisfying the desired priority over the entire system by processing 601 (see FIG. 3) such as communication and exclusive control between the clients 200, 300, 400, the logic becomes complicated. Therefore, there is a problem that the independence between the clients is lost, and the processing involved in scheduling becomes complicated and the overhead increases.

The present invention has been proposed to solve the above problems, and an object of the present invention is to provide a plurality of shared I / O devices for shared I / O devices without losing independence of each client and without increasing overhead. Another object of the present invention is to provide a method of controlling a shared input / output device, which can properly control the priority order of the input / output requests of the system over the entire system.

(Means for Solving the Problems) In order to solve the above problems, the present invention outputs the only input / output request having the highest priority at a point in time and a new input is issued until the processing is completed. In the method of controlling a shared input / output device shared by a plurality of independent clients that do not output an output request, each of the clients has the highest priority among all the requests that are currently generated as the current request. When outputting to a shared input / output device, the priority for the request and the priority for the request for the next output of the input / output request in the client are added to the current request to perform the shared input / output. Set in the device, the shared I / O device refers to all priorities set in addition to the request from each client, and the priority of the current request and the next output are scheduled. The priority of the completed I / O request is executed after the I / O processing corresponding to the current request corresponding to the highest priority of the requests to be performed is executed and the I / O processing is completed. Except for the priority, all the priorities added to the request from each client are referenced again, and the highest priority is added to the completed I / O request,
If the output of the I / O request is the priority of the scheduled request,
By waiting for the output of the next input / output request from the corresponding client and executing the input / output processing corresponding to the request, the input / output processing corresponding to the request with the highest priority is executed.

Note that while the shared I / O device is executing the I / O processing corresponding to a certain request, it is already set in the shared I / O device due to the change in the request that is scheduled to output the next I / O request that occurred in a certain client. When it becomes necessary to change the priority order of the existing priority, it is desirable that the priority of the next request can be reset to the shared I / O device.

(Operation) (1) In the present invention, each client shares the current request of the client, the priority of this request, and the priority of the request when the next request has already occurred. Passed to output device.

(2) Then, the shared input / output device compares the “priority of the current request” with the “priority of the next request” among the information of (1) passed from all the clients. Determines the highest priority one, and regardless of whether the highest priority one is the "current request priority" or the "next request priority", The current request included in the I / O request is output to the I / O medium for processing.

At this stage, if "priority of next request">"priority of current request" (priority of next request is higher than priority of current request), current request Is pushed out according to the "priority of the next request" and input / output is performed. When the I / O is completed, if the “priority of the next request” is also higher than the “priority of the current request” and the “priority of the next request” output by another client, Waits for the next I / O request from the client to be output after the completion of the I / O processing of the current request, and then inputs / outputs it to / from the I / O medium.

(2-1) When only one request (that is, only the current request) is generated in a specific client, the "priority of the next request" is not passed to the shared input / output device.
In claim 1, the current request is compared with the priority from another client by its own "priority of the current request", and the I / O request related to the pushing operation and the next request described above is compared. No output wait occurs.

(2-2) In an input / output request when a plurality of requests have already occurred in a specific client, since the requests are arranged in the order of priority within the client, “priority of current request”> “ The priority of the next request "(the priority of the current request is higher than the priority of the next request), and in claim 1, the pushing operation does not occur, but the input / output related to the next request Waiting for request output may occur.
However, in this waiting state, the input / output of the current request has already been completed, and the corresponding client always outputs the next request as a new input / output request, so there is no permanent waiting state. .

(2-3) Before the output of the input / output request to the shared input / output device and the processing of the shared input / output device is completed in (2-1) or (2-2) described above, the “current” is newly added. When a request having a priority higher than the “request priority” is generated, the request becomes the “next request” by rearranging the requests in the client in the order of priority. This is because the current request has already been output to the shared I / O device side, and it is generally unknown when the priority becomes the maximum and the I / O processing starts, and it is attempted to recognize this accurately. Then, exclusive control or the like is required between the client and the shared input / output device, which complicates the processing and lowers the performance. Therefore, this is a method for simplifying the processing.

The priority of the "next request" is passed to the shared input / output device by the resetting process of claim 2, and when the priority becomes higher than the priority from another client, Push-out operation occurs.

When a push operation occurs and the I / O processing corresponding to the I / O request is completed, the client that generated the I / O request replaces the previous "next request" with the "current request".
I / O request is newly output as, but the fact that the extrusion has occurred is that the "priority of the next request" of that client, that is, the new "priority of the current request" is the same for all clients. It indicates that it is the highest priority output by, so the shared I / O device waits for the I / O request to perform I / O to the I / O medium. Also, in this waiting state, the input / output of the current request has already been completed, and the corresponding client always outputs the next request as a new input / output request, so there is no permanent waiting state. .

(Example) Hereinafter, an example of the present invention will be described with reference to the drawings.

FIG. 1 shows the configuration of this embodiment.
Reference numeral 0 is a shared input / output device, 200, 300, and 400 are clients such as microprocessors, and 500 is an input / output medium of the shared input / output device 100. Then, the current request 201, 301, 401 is generated in each client 200, 300, 400, and thereafter, the request 202, 3
It is assumed that 02 and 402 are scheduled to be generated, respectively, and these requests are output to the shared input / output device 100 as input / output requests 210, 310, and 410. Here, the priority of the requests 201, 301, 401, 202, 302, 402 is the request 201> 202
>401>402>301> 302 (the left one has a higher priority).

When the client 200 makes the current request 201 to the shared I / O device 100, the client 200 adds the priority 21 of the request 201 and the next request 2 in the client 200.
If 02 has already occurred, the priority 22 associated with the request 202 having the next highest priority is added to the request 201.
This process is the same for the other clients 300 and 400, and priorities 31, 32 and 41, 42 are added when the requests 301, 401 of the clients 300, 400 are output.

In the processing in the shared input / output device 100 in this case, first, the priorities 21, 31, 41, 41 for the requests 201, 301, 401, 202, 302, 402
22, 32, 42 are compared, and the current request 201 having the highest priority among them is selected by the scheduler 101 and executed. That is, the input / output device 102 for the input / output medium 500 is executed according to the request 201.

Next, when the execution of the request 201 is completed and reschedule is performed, the priorities 31, 41, 22, 32, 42 of the remaining requests 301, 401, 202, 302, 402 are first compared. If the highest priority among these is the priority 22 of the request 202 scheduled to output the input / output request, which is added to the completed request 201, the corresponding client 2
Wait for the next request 202 to be output from 00, then request 2
The input / output processing corresponding to 02 is executed.

It should be noted that when only one request (that is, only the current request 201) is generated in a specific client, for example, 200, the “priority of the next request” is not sent to the shared input / output device 100. Current request 2 for I / O device 100
Since 01 is compared with the priority from other clients by its own priority 21, the pushing operation described above and the waiting for output of the I / O request for the next request do not occur.

As in the embodiment, in an input / output request when a plurality of requests 201 and 22 have already occurred in a specific client, for example, 200, the requests are arranged in the priority order in the client 200. The priority 21 of the request 201 is higher than the priority 22 of the next request 202, and the pushing operation described above does not occur, but when the process corresponding to the current request 201 is completed,
Waiting for output of an I / O request for the next request 202 may occur. However, in this waiting state, the input / output of the current request 201 has already been completed, and the client 200 must make sure that the next request 202
Is output as a new input / output request, so there is no permanent waiting state.

Hereinafter, in the same manner, each time a request is executed, the remaining priorities in the shared I / O device 100 are compared, attention is paid to the request with the highest priority, and the above is waited until the request is output. Repeat the above process.

As a result, even if there is a time lag between the execution of the highest priority request and the output of another request having the next highest priority, the request of a lower priority is executed first. Therefore, proper priority control can be performed through all the clients 200, 300, 400. The process in the above-described embodiment will be described with reference to FIG. 4. As indicated by a at the time of FIG. 4, the shared input / output device 100 requests the client 200 to execute the next request having the priority value of 9. It can be expected that the request 202, 2B (9), which is the request 202, will be output as the current request 201 of the client 200. Waiting for the output of the request, the next request of the client 300 in b is transmitted. Request 3
Input / output processing for the request 202 can be performed prior to 02 (3B (5)). As a result, during the input-output processing time
The processing is executed in the order of (9) → 3B (5).
The legend shown in FIG. 5 is common to FIG.

Next, in FIG. 1, the priority order of requests 201,202,301,302,401,402 of the clients 200,300,400 to the shared input device 100 is 201>301>401>202>302> 402 (priority 21>31>41>22>32> 42). Yes, each client
Clients 200, 300, 400 make requests in the same way as the previous example, and while the shared I / O device 100 is executing processing for request 201, the client
Requests with priority 23 are newly scheduled within 200, and the priority of all remaining requests is 203>301>401>202>302> 402 (priority
23>31>41>22>32> 42) will be described below.

In this case, the client 200, as shown in FIG.
For the shared I / O device 100, the priority is reset 220 for the request 203 having the next highest priority after the request 201. That is,
The priority 22 of the next request 202 until then, the request 203
Change to priority 23.

In other words, the request 203 becomes the “next request” by rearranging the requests in the client 200 in order of priority, and
When the priority 23 of the request 203 becomes higher than the priority of the request of another client (for example, the priority 31 of the request 301) by the resetting process of the “priority of the next request”, the push state is changed. Occur.

When the push-out state occurs and the input / output processing corresponding to the input / output request 201 is completed, the client 200 that generated the input / output request 201 newly sets the “next request” up to that point as the “current request”. I / O request 203 is output, but the fact that extrusion has occurred is that the client
The 200 previous "next request priority" or the new "current request priority" indicates that it is the highest priority output by all clients, so the shared input The output device 100 waits for the input / output request 203 to be output and then inputs / outputs to / from the input / output medium 500. Further, even in the waiting state for the output of this input / output request 203,
Since the input / output of the current request 201 has already been completed, and the corresponding client 200 always outputs the next request 203 as a new input / output request, it does not enter a permanent waiting state. By doing so, it is possible to perform legitimate scheduling based on the priority of each request, regardless of changes in the circumstances of each client.

The present invention is of course applicable to not only a multiprocessor system having a common bus but also a system in which a plurality of independent microprocessors are connected via a communication line.

(Effects of the Invention) As described above, according to the present invention, the shared input / output device performs scheduling by referring to all priorities of the current request and the next-order request generated from each client, and if necessary, Since it is possible to re-set the priority level, it is possible to simply realize a proper priority order control over the entire system without requiring communication between clients or exclusive control as in the related art.

Therefore, there is an effect that the independence between the clients is not lost and the overhead of the shared input / output control can be reduced.

[Brief description of drawings]

FIG. 1 is a block diagram of an input / output device sharing system to which an embodiment of the present invention is applied, FIG. 2 is an explanatory diagram when a new request occurs in a client, and FIG. FIG. 4 is a configuration diagram of an output device sharing system, FIG. 4 is an explanatory diagram of processing in the embodiment of the present invention, and FIG. 5 is an explanatory diagram of processing in a conventional example. 100 …… Shared I / O device 200,300,400 …… Client 201,202,203,301,302,401,402 …… Request 210,310,410 …… I / O request 21,22,23,31,32,41,42 …… Priority

Claims (2)

[Claims] 1. A plurality of independent devices (hereinafter referred to as clients) that individually output the only I / O request having the highest priority at a time point and do not output a new I / O request until the processing is completed. In the shared shared I / O device control method, each client outputs the highest priority request among all currently generated requests to the shared I / O device as a current request. The priority of the request and the priority of the request for which the output of the next I / O request is scheduled in the client are added to the current request and set in the shared I / O device. Refers to all priorities set in addition to the request from each client, and has the highest priority among the priority of the current request and the priority of the request to be output next. The I / O processing corresponding to the current request corresponding to the high I / O request is executed, and after the I / O processing is completed, the priority of the completed I / O request is removed and the request is added again to the request from each client. If all the priorities set by the above are referred to, and the highest priority is the priority of the scheduled output of the I / O request, which is added to the completed I / O request, , Waiting for the output of the next I / O request from the corresponding client, and executing the I / O processing corresponding to the request, thereby performing the I / O processing corresponding to the request with the highest priority. Shared I / O device control method. " 2. A shared I / O device, while executing an input / output process corresponding to a certain request, by changing a request which is generated in a certain client and is scheduled to output the next input / output request,
When the priority of the priority already set in the shared input / output device needs to be changed, the priority of the next request can be reset in the shared input / output device.
A method for controlling a shared input / output device according to the item.