JPH07271704A - Device controller - Google Patents

Abstract

PURPOSE:To provide the device controller which can control plural different kinds of processor type device only by expanding one driving program on a memory. CONSTITUTION:The device controller, which controls the processor type devices 103 and 106 connected to a SCSI bus 102, is provided with control object prescribing means 130 and 119 which prescribe a device to be controlled and a set of device files corresponding to the device, a device recognizing means 129 which inspect whether or not a processor type device is included in the device to be controlled based on the device information obtained from the processor type device. When the device is included in the set, a device file is generated. Since, the device to be controlled is prescribed with the set, the need for a program which recognizes respective devices individually is eliminated and plural different kinds of processor type devices can be controlled by one driving program, thereby saving memory resources.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a device control device for controlling a processor type device having a SCSI interface, and more particularly to a device control apparatus capable of controlling a plurality of different types of processor type devices with one drive program. is there.

[0002]

2. Description of the Related Art In recent years, SCSI has attracted attention as a standard interface for personal computers and workstations.
(Small Computer System In
The interface is a standard interface for connecting a small computer and a peripheral device, which is approved by ANSI (American National Standards Institute). A maximum of eight devices can be connected to the bus that is interfaced by SCSI, and each device is given an identification number of 0 to 7. Each device may be an initiator or a target when transferring data on the bus. An initiator is a device equipped with a function of selecting a required device and giving a command to control the device.
The target is a device having a function of executing an instruction command under the control of the initiator. When selecting a target, the initiator has a logical unit number (LUN: Logical) attached to a child device called a logical unit of each target device.
(Unit Number) is specified.

The types of devices defined by the SCSI standard include direct access devices such as hard disks and sequential access devices such as magnetic tapes.
There are devices and processor type devices, and in addition, there are printer devices, scanner devices, write-once devices, CD-ROM devices, optical storage devices such as MO disks, medium exchange devices, and communication devices. The standard command defined for each device type is defined as the command sent from the initiator.

The processor type device is a target having a function as a computer device, and according to the command received from the initiator, data packet transmission / reception, attribute information transfer, sense data acquisition, and existence confirmation of the device on the bus are performed. Do.

The commands used between the initiator and the processor type device will be described below. The initiator issues a RECEIVE command to the target when receiving data from the target, and issues a SEND command when transmitting data to the target. Since the content of the information sent by the SEND command and the RECEIVE command is not specified, the program for controlling the drive of the processor type device unit can correspond to various processor type device units.

Further, when the initiator acquires the attribute information of the device, the initiator INQUIRs the target.
Issue a Y command. In response to this, the attribute information of the device is notified from the target as INQUIRY data. If the command executed previously is an error, the R
Issue the EQUEST SENSE command and acquire the reason for the abnormality as SENSE data from the target. Also, the initiator will check if there are any devices on the bus.
When checking whether it is in a usable state, a TEST UNIT READY command is issued to the target.

Next, the structure of an apparatus for controlling a SCSI processor type device will be described. FIG. 10 is a block diagram of a conventional SCSI processor type device control device in the Solaris 2.2 system of Sunsoft Corporation. In FIG. 10, a user process 1001 transfers data with the device by issuing a system call. Reference numeral 1002 denotes a Solaris 2.2 operating system, which is an OS of Sunsoft Corp. of the United States of America, which effectively allocates hardware resources to a plurality of user processes. A magnetic disk type target driver 1003 controls driving of the hard disk type device 1008, and executes a system call issued by a user process 1001 to the SCSI for the hard disk type device.
It is converted into a command, and the host adapter driver 1006 is requested to execute the command. A magnetic tape type target driver 1004 controls the drive of the magnetic tape type device 1009, converts the system call issued by the user process 1001 into a SCSI command for the magnetic tape type device, and executes the command by the host 1006. Ask the adapter driver. A processor device type target driver 1005 controls driving of the processor device type device 1010, converts a system call issued by a user process 1001 into a SCSI command for the processor device type device, and executes the command by a host 1006. Ask the adapter driver. 100
A host adapter driver 6 issues a SCSI command requested by each target driver to each device. Reference numeral 1007 is a SCSI bus for connecting the computer main body and peripheral devices.

Next, a procedure for recognizing a SCSI processor type device will be described. FIG. 11 is a flow of a procedure for recognizing a conventional SCSI processor type device, which is executed by the processor type device target driver 1005.

Step 1101; the processor-type device target driver 1005 uses the TEST UNIT
Issue a READY command to verify that the devices on the bus are ready for use.

Step 1102; According to the inspection result,
If the device cannot be used, step 1109; The error code is set to indicate that the device is defective, and the processing ends in error.

Step 1103; When the device is available, the INQUIRY command is issued to display the device type, which is the device information, the vendor, the product ID, and other unique information of the device.
Obtained as RY data.

Step 1104: It is checked whether or not this execution command has ended normally, and if there is an error, Step 11
10; The error code is set to indicate that there is no response to the inquiry, and the processing ends in error.

Step 1105: When this execution command is completed normally, it is checked whether the device type is a processor type device based on the acquired device information. If it is not a processor type device, step 1111; The error code is set to the effect that it is not a processor type device, and the processing ends in error.

Step 1106: If it is a processor type device, it is checked based on the device information whether or not the product ID corresponds to a control target device fixed in advance in the program of the processor type device target driver. If not, step 1112; it is set that the device is not the corresponding device, and the process ends in error.

Step 1107: If the device is a control target device, a work area is secured, and step 1108; a device file having a name corresponding to the control target device is created, and the process ends normally.

This device file is an access point used when the user process 1001 controls the device, and is created in the Solaris 2.2 operating system.

Next, a data transfer control procedure of the SCSI processor type device controller will be described. FIG. 12 is a flow chart showing a data transfer control procedure of the conventional SCSI processor type device control apparatus, which is called when the user process 101 of FIG. 10 issues a system call such as read or write to the processor type device apparatus 1010. It

Step 1201; Get the resources needed to execute the SCSI command. Step 1202: The system call issued by the user process to request the processor type device for processing is converted into a SCSI command for the processor type device unit. For example, in case of read system call, RE
The CIVE command, the SEND command for the write system call, the ioctl system call differ depending on the instruction command, and the TEST UNIT READY command for the command to confirm the existence of the device.
When it is a device information acquisition command, it is converted into an INQUIRY command.

Step 1203: The host adapter driver 1006 is requested to execute the SCSI command created in Step 1202, and Step 1204; the command execution completion waiting state is entered. When execution is complete, step 1
205; analyze the execution result of the SCSI command,
Step 1206: It is checked whether or not the result of the analysis is normal, and if there is no abnormality, the operation ends normally. If an abnormality is detected, Step 1207; an error code is set and the operation ends in error.

[0020]

However, the conventional device control apparatus has the following problems. First, since the product ID of the device to be controlled is predetermined in the program of the processor-type device target driver 1005, when controlling the driving of a plurality of types of processor-type devices, each device is controlled individually. A drive program with a dedicated recognition processing unit is required.
Therefore, even if the data transfer control processing unit adopting the procedure shown in FIG. 12 can be commonly used for a plurality of devices, the recognition processing unit is redundantly expanded in the memory, and the memory resources are wasted. It had the problem of being exposed.

Second, when the device to be controlled is OEM, the product ID may be changed at the OEM destination according to the vendor of the device. If the product ID has been changed, the product ID recognition processing unit proceeds to step 110 of FIG.
Since the product ID “A” shown in FIG. 6 has to be changed sequentially, there is a problem that the maintenance is troublesome and the driving program is poor in versatility.

Thirdly, in the conventional system, when the control target is a SCSI device, the system call issued by the user process to the processor type device in step 1202 of FIG. 12 is converted into a SCSI command, and step 1
In step 203, the host adapter driver 1006 is requested to execute the command, and in step 1204, the command execution end wait state is entered. In this execution end wait state, the system call is interrupted even if a signal is received. I can't. Therefore, if the device does not receive the execution completion notification of the command, the system call does not end for many hours, so it is impossible for the user process to perform processing other than device control, such as pipe reading, in parallel. Become.

When the execution of the SCSI command is requested in step 1203 as a means for interrupting the running SCSI command after a lapse of a certain time and returning control to the user process, the host adapter driver 1006
There is a method of designating a timer for monitoring the completion of execution of the command. However, in this case, when the timer times out, all the devices are reset regardless of the state of the device, which may cause a problem in some devices.

For example, in an 8-channel FAX communication device (FEP device shown by 103 in FIG. 1) which will be described later, when the RECEIVE command is normally received from the initiator, the bus is disconnected and the bus is connected when the communication event occurs. Reconnection), the content of the communication event that occurred is transferred to the initiator as data, and the end status is similarly transferred to the initiator. However, since the time until the arrival of the next communication event cannot be limited within a fixed time, when the device is in the idle state, there is no communication event and the time until the end of the RECEIVE command cannot be predicted. In such a case, if the timer is used, the host adapter driver 100 can be executed during the execution of the command.
6 may reset the device. In that case, the device will be in a state after the power is turned on, and the processes that should be performed at the time of startup, such as restarting and setting the initial values of system parameters, are required, and all the processes that have been performed up to that point are wasted. A problem occurs.

For this reason, in the conventional apparatus, when the execution of a command is started, the command being executed is awaited for its end without being interrupted. Therefore, even if a signal is received, the system call cannot return, so that the user process has a problem that it cannot perform other processing such as pipe reading in parallel with device control.

The present invention solves these conventional problems and saves memory resources by controlling a plurality of different types of processor-type devices simply by developing a single drive program on the memory. It is an object of the present invention to provide a device control device capable of

It is another object of the present invention to provide a device control apparatus equipped with a highly versatile program that does not require a driver program for drive control to be changed each time even when device information is changed. I am trying.

Even if there is no command execution end notification from the device, the command being executed can be interrupted for a certain period of time to terminate the system call, and a device which can prevent the trouble caused thereby can be suppressed. The purpose is to provide a control device.

[0029]

Therefore, in the present invention, S
In a device control device for controlling a processor type device connected to a CSI bus, a control target defining means for defining a set of a device to be controlled and a device file corresponding to the device, and a device to be controlled from the processor type device. Device recognition means for inspecting whether this device is included in the devices to be controlled based on the device information, and is configured to create the device file when the device is included in the set. .

The controlled object defining means is composed of a variable defining a set of devices and their names, and a file for setting the contents of this variable.

Further, a timer for monitoring the completion of execution of the SCSI command, a timer value setting means for setting the timer value of the timer, and an S which is being executed when the timer times out.
Abort processing means for aborting (interrupting) the CSI command without resetting the target is provided.

Further, setting means for setting the type of event that needs to be processed and inspection means for inspecting the occurrence of this event are provided, and when the timer times out, the abort processing is performed only when this event has occurred. Causes the means to perform an abort process.

Further, an inspection means for inspecting the presence or absence of a device failure is provided, and after the abort processing means performs the abort processing, this inspection means is configured to inspect the device.

[0034]

Therefore, since the devices to be controlled are defined by a set, a program for individually recognizing the devices is not required, and one drive program controls a plurality of different types of processor type devices. It becomes possible to save memory resources.

Further, since the controlled object is defined by the variable and the content of the variable is defined by the file, the OEM
Even if the product ID is changed in advance, it can be dealt with by changing the content of the file, which simplifies the maintenance of the program and increases versatility.

Further, in the case where the timer for monitoring the completion of the execution of the SCSI command is provided, in the case of timeout, the abort processing means suspends the command being executed without resetting the device. As a result, control can be returned to the user process, and the user process can process other events. Also,
If it becomes necessary to restart the execution of the SCSI command, the device is not reset and can be restarted quickly.

With the means for setting the type of event, the abort process is executed only when the set event actually occurs in the case of a timeout, and the timer is restarted in other cases. Start. Therefore, the number of abort processing of the command being executed decreases,
The load on both the initiator and target can be reduced. Further, the continuation of command execution can be prioritized, and when an event having a priority higher than that can occur, the process can be executed immediately after a timeout.

Further, in the case where the device investigation means is provided, the time-out may be due to the device failure, or the device may be operating normally but there is no data to be sent to the initiator. Can be separated, and by notifying the user process of this determination result, the user process can
It is no longer necessary to issue a system call by itself to check the device status, and the load can be reduced.

[0039]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The configuration of a SCSI processor type device controller according to an embodiment of the present invention will be described with reference to the drawings.

FIG. 1 is a block diagram of a device for controlling the FEP device and the VRC device which are processor type device devices. In FIG. 1, reference numeral 101 denotes a host computer, which is an initiator having a function of giving a SCSI command to a target processor-type device unit to control it. The SCSI bus 102 includes a host computer 101, a FEP device 103, and a VR VR 106.
Connect to C device. The FEP device 103 is an 8-channel FAX communication device, and includes two logical units 104, 1
With 05. Logic unit 0 at 104 accepts the SEND command from the initiator and receives instructions and data from the initiator. Logical unit 1 of 105
It receives a RECEIVE command from the initiator and transfers a response and data to the initiator. 106
The VRC device is a device for rasterizing code information, and like the FEP device 103, it has two logical units 10.
7 and 108. Logic unit 0 at 107 accepts the SEND command from the initiator and receives instructions and data from the initiator. The logical unit 1 of 108 receives the RECEIVE command from the initiator and transfers a response and data to the initiator.
109 is an FEP control transmission process, which is a logical unit 0
115, which is the device file corresponding to (104)
A write system call is issued to / dev / fep0 of the device and the instruction and data are transferred to the FEP device 103. Reference numeral 110 denotes an FEP control reception process, which is a device file corresponding to the logical unit 1 (105).
A read system call is issued to 16 / dev / fep1, and a response and data from the FEP device 103 are received.

The processes 109 and 110 are connected by a pipe 111, which is a bidirectional data path, and perform data transfer with each other in order to synchronize the control of the FEP device 103. The FEP control transmission process 109
When transferring data to the FEP control reception process 110, a write system call is issued to the pipe 111. At that time, the FEP control transmission process 109
Issue an interrupt signal called SIGINT and
The P control reception process 110 is notified of the arrival of data. Upon receiving this signal, the FEP control reception process 110 issues a read system call to the pipe 111 and receives the data from the FEP control transmission process 109. FEP control reception process 110 to FE
Similar processing is performed in the case of data transfer in the reverse direction to the P control transmission process 109. Reference numeral 112 denotes a VRC control process, which is a device file corresponding to the logical unit 0 (107) and is wr for 117 / dev / vrc0.
ITE system call to transfer instructions and data to the VRC device 106, and also to the logical unit 1 (10
8 / d, which is a device file corresponding to 8)
It issues a read system call to ev / vrc1 and receives a response and data from the VRC device 106. Reference numeral 113 denotes a Solaris 2.2 operating system (hereinafter referred to as an OS), which is an OS of Sunsoft, Inc. in the United States, which effectively allocates hardware resources to a plurality of user processes.

The OS 113 has 114 file systems. The device file / dev / fep0 (115) on the file system 114 provides the FEP control transmission process 109 with an access point to the logical unit 0 (104) of the FEP device 103, and also / dev / The fep1 (116) provides the FEP control reception process 110 with an access point to the logical unit 1 (105) of the FEP device 103, and / dev / vrc0 (117) provides the VRC control process 112. To provide an access point to the logical unit 0 (107) of the VRC device 106, and /
dev / vrc1 (118) is the VRC control process 1
12 to the logical unit 1 (1
08) to provide access points. 119 of / e
The tc / system file defines system parameter information initial values inside the OS 113.
Reads the file at startup and sets the specified initial values for the specified variables.

Reference numeral 120 denotes a processor type device target driver sprc, which is used for the FEP device 103 and VRC.
The drive control of the device 106 is performed, and the process 109,
The system calls issued by 110 and 112 are converted into SCSI commands for the processor type device unit, and the host adapter driver 131 is requested to execute the commands. The host adapter driver 131 issues the SCSI command requested by the sprc driver 120 to each of the device units 103 and 106.

The sprc driver 120 has a timer value setting process 121, and an interface for designating the value of the timer variable 122 is the process 109,
It is provided to 110 and 112. timer variable 12
2 is 1 for each logical unit 104, 105, 107, 108 of the FEP device 103 and VRC device 106.
SCs that exist one after another and are issued by the sprc driver 120 to the logical units 104, 105, 107, 108.
The SI command execution end waiting time is set. Further, the sprc driver 120 has a user process ID setting process 123, and an interface that can specify the value of the process variable 124 is the process 1
09, 110, 112. procid
There is one variable 124 for each logical unit 104, 105, 107, 108 of the FEP device 103 and VRC device 106, and sets the process ID of the process (109 to 112) that controls each logical unit. It is a thing. Also, the sprc driver 120
25 signal number setting processing, 126 sign
An interface is provided to the processes 109, 110, 112 that allows the value of the o variable to be specified. This si
The gno variable 126 is the FEP device 103, the VRC device 1
06 logical units 104, 105, 107, 108
There is one for each, and the process (109 to 112) that controls each logical unit sets an event that requires processing by a signal number. Furthermore, the sprc driver 120 has 127 abort processing, and interrupts (aborts) the SCSI command being executed without resetting the target. Reference numeral 128 denotes a data transfer process, which includes the FEP device 103 and the VRC device 10.
Data is transferred to 6. Reference numeral 129 is a device recognition process, which checks whether the connected devices 103 and 106 are processor-type devices, or whether those devices correspond to the devices specified by the dev_set variable of 130. Just dev_
The set variable 130 is / etc / system of the OS 113.
Character string specified in em file 119
A device (collection of devices) that is a pointer that indicates an actual control target is represented by a character string in this / etc / system file 119.

(First Embodiment) A first embodiment of the present invention will be described below with reference to the drawings. 2 is / e
A part of the tc / system file 119 is shown, and the processor type device target driver sprc is shown.
It represents a value for initializing the contents of the dev_set variable 130 which is an internal variable of 120. OS113 is sp
When the rc driver 120 is activated, the dev_set variable 130, which is a pointer to a character, is initialized to point to the character string shown in FIG.

The character string shown in FIG. 2 is the sprc driver 1
The device 20 controls the drive and the device file name thereof. The product ID of the device that the sprc driver 120 controls the drive is on the right side of “:”, and the device file name of the sprc device 120 is on the left side. A plurality of devices to be controlled by the sprc driver 120 can be defined by defining them as one set and separating the plurality of sets by separating them with commas. In FIG. 2, the product ID is FEP and the device file name is fep.
And a product ID of VRC and a device file name of vrc are defined as control target devices.

FIG. 3 is a flow of a procedure for recognizing a SCSI processor type device, which is device recognition processing 129.
Of the processor type device target driver sprc.

Step 301; TEST UNIT R
By issuing the EADY command, it is checked whether the device exists on the bus or is in a usable state.

Step 302; If the inspection result indicates that the device cannot be used, Step 311; The error code is set to indicate that the device is defective, and the processing ends with an error.

Step 303; if available,
By issuing the INQUIRY command, the device type that is the device information, the vendor that is the manufacturer, the product ID, and other unique information of the device are acquired as INQUIRY data.

Step 304: It is checked whether or not the execution command has ended normally. Step 312: If an error occurs, the error code is set to indicate that there is no response to the inquiry, and the processing ends in error.

Step 305: When the processing is normally completed, it is checked whether the device type is a processor type device based on the device information. If it is not a processor type device type, step 313; the fact that it is not a processor type device is set in the error code, and the processing ends in error.

Step 306: When the device is a processor type device, the character string up to the next comma in the character string pointed to by the dev_set variable 130 is cut out, and the product ID of the next device to be controlled and its device file name are acquired. Then, the local variables pid and dname are set respectively.

Step 307; Judge the result of the variable setting, and if there is no device which becomes the next element, step 3
14: The error code is set to indicate that there is no corresponding device, and the processing ends with an error.

Step 308: If there is a device to be the next element, it is checked based on the device information whether the product ID matches the pid variable. If they do not match, the process returns to step 306. If they match, step 309; a work area is secured, step 310; a device file is created based on the dname variable, and the process ends normally.

As described above, in the present embodiment, the dev_set variable 13 that defines the set of devices to be controlled
0 is set, the actual content of the variable 130 is described in the / etc / system file 119, and unique information such as the product ID of the SCSI device is acquired by the device recognition processing 129, and the product ID is included in the variable. If it is included, a device file with the device name defined by the above variable is created, so that a single type of heterogeneous processor type device with the same data transfer control method is driven. It can be controlled by the control program, and the system memory can be used efficiently. Even if the device to be controlled is OEM and the product ID is changed, the content of the dev_set variable 130 is described in the / etc / system file 119. Therefore, the content of this file is the product ID of the OEM destination.
It is not necessary to change the source of the drive control program simply by changing to, maintenance can be facilitated and the versatility of the program can be increased.

(Second Embodiment) A second embodiment of the present invention will be described below with reference to the drawings. FIG. 4 is a flow showing the processing procedure of the timer value setting processing 121.
The timer value setting process 121 is the process 109, 1 of FIG.
For 10, 112 (hereinafter referred to as control process), i
It provides a SETTIMER instruction interface for the octl system call, and allows the control process to set the value of the timer variable 122, which is a timer value that monitors the end of execution of a SCSI command. When executing the SETTIMER instruction, the control process specifies the file descriptor of the device file corresponding to the logical unit in charge of control and the timer value to be set, as parameters of the ioctl system call. The file descriptor is a value returned as a return value when the device file is opened by issuing the open system call.

Step 401; It is checked whether the input timer value is a specifiable value. Step 402; If the timer value is an unspecifiable value, Step 404; The error code indicates a timer value error. Set a message and end with an error.

Step 403: When the timer value is normal, the timer value is stored in the timer variable 122 provided for the logical unit corresponding to the file descriptor, and the ioctl system call is terminated normally.

FIG. 5 is a data transfer control flow of the sprc driver 120 using the timer variable 122 set by the timer value setting process 121, and shows the contents of 128 data transfer process. This process is called when the control process issues a system call such as read or write to the FEP device 103 and the VRC device 106.

Step 501: Acquire the resources required to execute the SCSI command. Step 502; The system call issued by the control process to request the FEP device 103 or the VRC device 106 for processing is converted into a SCSI command for the processor type device device. For example, REC for read system call
Convert to EIVE command or SEND command in case of write system call, and in case of ioctl system call, according to the instruction content, if there is an instruction to confirm the existence of the device, TEST UNIT REA
For the DY command, and for the device information acquisition command,
Convert to INQUIRY command respectively. When issuing these system calls, the control process issues the file descriptor of the device file corresponding to the logical unit in charge of control, the data reception buffer and its size in the case of the read system call, and the write command.
In the case of a system call, the data transmission buffer and its size are specified as parameters. This file descriptor is the corresponding logical unit number, and
The buffer size is the number of transfer bytes
It is reflected in the SCSI command of IVE and SEND.

Step 503: A timer is started based on the timer variable 122 provided for the logical unit corresponding to the file descriptor.

Step 504: The host adapter driver 1 executes the SCSI command created in step 502.
31 and step 505; the command execution waiting state is entered. When the execution of the command is completed, step 506; It is checked whether the timer is in the ended state due to timeout. In the case of time-out, the abort process is executed by the abort process 127 in step 507, and step 508 is set in the error code indicating that the time is out, and the process ends in error.

Step 506; If the end status of the command is not timeout, step 509; SCS
The I command execution result is analyzed, step 510;
It is checked whether the result of the analysis is normal, and if normal, the process ends normally. If an abnormality is detected, step 51
1; Sets an error code and ends with an error.

As described above, in this embodiment, the timer value setting process 121 sets the timer value of the timer variable 122 for monitoring the completion of execution of the SCSI command, and when the timer times out, the device (103, 106) Aborts the SCSI command that is being executed without resetting. Therefore, even if there is no response from the device side to the execution of the command, control can be returned from the system call to the user process after a certain period of time, and the control process can process other events such as 111 pipe reading. It becomes possible to do it.

(Third Embodiment) A third embodiment of the present invention will be described below with reference to the drawings. FIG. 6 is a flow showing the processing procedure of the user process ID setting processing 123. This process is executed by the io for the control process.
Provides the SETPROCID command interface for the ctl system call so that the controlling process can
It is possible to notify D to the sprc driver 120. When executing the SETPROCID instruction, the control process specifies, as parameters of the ioctl system call, the file descriptor of the device file corresponding to the logical unit in charge of control and its own process ID. The file descriptor is the device file
This is the value returned as the return value when opened by issuing the pen system call.

Step 601: Specified process ID
Check if is actually present, step 60
2; Step 60 if the process ID does not exist
4; The process error is set to the error code, and the process ends with an error.

Step 603: When the process ID exists, the specified process ID is stored in the process variable 124 provided for the logical unit corresponding to the file descriptor, and the ioctl system call is normally terminated. .

FIG. 7 is a flow showing the contents of the signal number setting processing 125. This signal number is SCS
When the execution of the I command is not completed within a certain period of time, the type of processing that needs to be executed instead is represented by a number. This signal number setting process is executed by the control process, SETUP signal of ioctl system call.
It provides an O instruction interface, and the control process specifies the number of the signal that needs to be processed by the sprc driver 12
0 can be notified. The control process is SE
When executing the TSIGNO instruction, the file descriptor of the device file corresponding to the logical unit in charge of control and the signal number that requires processing are specified as parameters of the ioctl system call.

Step 701: The value of the designated signal number is inspected. Step 702; If this signal number is an invalid value, Step 704; The error code is set to indicate that a signal number error has occurred. End with an error.

Step 703: When the signal number is normal, the signal number is stored in the signo variable 126 provided for the logical unit corresponding to the file descriptor, and the ioctl system call is normally terminated.

FIG. 8 shows the user process ID setting process 12
3 is a data transfer control flow of the sprc driver 120 using the procid variable 124 set by No. 3 and the signo variable 126 set by the signal number setting process 125, and represents the contents of the data transfer process 128. In the processing shown in this flow, the control process is executed by the FEP device 103 or the VRC device 106 to
Called when a system call such as ad or write is issued.

In this flow, when the timer for monitoring the execution of the SCSI command times out, it is checked whether or not the processing request of the preset signal number is actually generated, and only when it is generated, the command is processed. Will be aborted.

The procedure from step 801 to step 805 is the same as the operation from step 501 to step 505 shown in the second embodiment, and the resources necessary for executing the SCSI command are acquired and received from the control process. The system call is converted into a SCSI command, a timer for monitoring the completion of execution of the command is started, the host adapter driver 131 is requested to execute the command, and the completion of execution is awaited.

Step 806; In step 803, the control process checks whether the timer started based on the timer variable 122 provided for the logical unit in charge of control has timed out. In the case of time-out, step 807; the process ID of the control process is acquired by the process variable 124 provided for the logical unit, and step 808;
Signo variable 1 provided for the logic unit
26, the control process having the process ID obtains the number of the signal that needs to be processed.

Step 809; Check if the signal is actually occurring for the process.
If it has occurred, step 810; abort processing 1
The abort process is executed according to 27, and step 811; the error code is set to timeout, and the process ends in error.

Step 812: If the signal is not generated, the timer is restarted and the process returns to step 805.

Step 813; If the termination status of the SCSI command is not timeout in step 806, the SCSI command execution result is analyzed. Step 814; If no abnormality is detected as a result of this analysis,
If the processing ends normally and an abnormality is detected, step 81
5: Set an error code and end with an error.

As described above, in this embodiment, the process ID
The setting process 123 sets the process ID of the control process, the signal number setting process 125 sets the number of the signal that the process needs to process, and the timer value setting process 125 sets the timer to set the SCS.
Start executing the I command. When the command execution end monitoring timer has timed out, the control means checks whether or not the above signal is generated, and if only the signal is generated, abort processing is performed and control is returned to the user process. Let After returning, the process can perform other processing to be processed such as reading the pipe of 111 in response to the generation of the signal.

As described above, in the third embodiment, unlike the second embodiment, the abort process is executed only when the event requiring the process occurs. Therefore, the device reduces the number of times the abort process is performed and the load is reduced. It

(Embodiment 4) In the fourth embodiment of the present invention,
In the data transfer control, a procedure for investigating the cause of the timeout of the timer for monitoring the completion of execution of the SCSI command is provided. This data transfer control flow is
It is shown in FIG. In FIG. 9, operations from steps 901 to 910 from acquisition of resources required to execute a SCSI command to execution of abort processing, and from step 915 when execution of a command ends without a timeout The operation up to step 917 is the same as in the case of the data transfer control of the third embodiment (FIG. 8).

Step 911; After executing the abort process, issue a TEST UNIT READY command to check whether or not the device on the bus is in a usable state. Step 912; As a result of this check, the device is not used. If it is possible, step 913; The error code is set to indicate that the device is in failure, and the processing ends in error.

In step 912, if the inspection result is normal, there is no data to be transferred to the control process, so execution is not ended, and it is judged that a time-out has been reached and the processing is normally ended.

As described above, in this embodiment, the SC being executed is
After the abort process is executed for the SI command, the state of the device (103, 106) is inspected, and the cause of the time-out of the timer for monitoring the completion of the execution of the SCSI command is investigated, and the device is unusable due to device failure or the like. The state is divided into a state in which the device operation ends normally, but there is no data to be sent to the initiator, and the device is in a waiting state for completion of execution of a SCSI command.

By notifying the user process of this determination result, the user process does not need to issue a system call to check the state of the device. Therefore, the processing in the user process can be simplified,
The system load can be reduced.

[0086]

As is apparent from the above description of the embodiments, in the device control apparatus of the present invention, a set of devices to be controlled is defined, and when a device is included in the set, it corresponds to the device. Since a device file with a name is created, one drive program can control a plurality of different types of processor-type devices that share the same data transfer control method, saving system memory. .

Further, since a set of devices to be controlled and a device file name thereof are defined by variables and a file for defining the variables is separately provided, it is possible to change the product ID of the device at the OEM destination. However, this can be dealt with only by changing the file, and there is no need to change the driver program for drive control. Therefore, maintenance is simplified and the versatility of the program can be increased.

Also, the completion of the execution of the SCSI command is monitored by the timer, and when the timer times out, the SCSI command being executed is aborted without resetting the target. Therefore, the response to the execution of the command is the device side. Even if it does not occur, the control capability can be restored to the user process after a certain time has elapsed, and the user process can perform processing for another event. Further, even when the execution of the SCSI command is restarted, the device is not reset, so that the restart can be promptly performed.

Since the abort process is executed only when the event of interest occurs, the number of abort processes is reduced while ensuring the opportunity to process the necessary event quickly. It is possible to reduce the load on the initiator and the target.

Further, after executing the abort processing, the cause of the time-out is investigated, and it may be due to a device failure.
Alternatively, the device is operating normally, but since it has timed out because there is no data to send to the initiator, the user process must issue a system call to check the device status. Can be reduced and the load can be reduced.

[Brief description of drawings]

FIG. 1 is a block diagram showing the configuration of a device control apparatus according to an embodiment of the present invention,

FIG. 2 is a diagram showing a part of an / etc / system file in the first embodiment,

FIG. 3 is a flow chart showing a SCSI processor type device recognition procedure according to the first embodiment;

FIG. 4 is a flowchart showing a procedure of timer value setting processing in the second embodiment;

FIG. 5 is a flowchart showing a transfer control operation to a processor type device according to the second embodiment;

FIG. 6 is a flowchart showing the procedure of user process ID setting processing in the third embodiment;

FIG. 7 is a flow chart showing a procedure of signal number setting processing in Embodiment 3;

FIG. 8 is a flowchart showing a transfer control operation to a processor type device in the third embodiment;

FIG. 9 is a flowchart showing a transfer control operation to a processor type device in the fourth embodiment;

FIG. 10 is a block diagram showing the configuration of a conventional SCSI processor type device control device;

FIG. 11 is a flowchart showing a processor type device recognition procedure in a conventional control device;

FIG. 12 is a flowchart showing a transfer control operation to a processor type device in a conventional control device.

[Explanation of symbols]

 101 Host Computer 102 SCSI Bus 103 FEP Device 104 Logical Unit 0 105 Logical Unit 1 106 VRC Device 107 Logical Unit 0 108 Logical Unit 1 109 FEP Control Send Process 110 FEP Control Receive Process 111 Pipe 112 VRC Control Process 113 Solaris 2.2 Operating System System 114 File system 115 / dev / fep0 device file 116 / dev / fep1 device file 117 / dev / vrc0 device file 118 / dev / vrc1 device file 119 / etc / system file 120 Processor type device target driver 121 Timer value setting process 122 timer variable 123 user process ID setting process 124 p ocid variable 125 signal number setting setting process 126 signo variable 127 abort process 128 data transfer process 129 device recognition process 130 dev_set variable 131 host adapter driver 1001 user pro process 1002 Solaris 2.2 operating system 1003 magnetic disk type target driver 1004 magnetic tape type Target driver 1005 Processor type device Target driver 1006 Host adapter driver 1007 SCSI bus 1008 Hard disk type device 1009 Magnetic tape type device 1010 Processor type device device

Claims (5)

[Claims] 1. A device control apparatus for controlling a processor-type device connected to a SCSI bus, and a control target defining means for defining a set of a device to be controlled and a device file corresponding to the device. A device recognition unit that checks whether or not the device is included in a device to be controlled based on device information obtained from the processor-type device, and creates the device file when the device is included in the set A device control device characterized by the above. 2. The device according to claim 1, wherein the controlled object defining unit is configured by a variable that defines a set of a device and its name, and a file that sets the content of the variable. Control device. 3. A device control device for controlling a processor type device connected to a SCSI bus, comprising: a timer for monitoring the completion of execution of a SCSI command; and a timer value setting means for setting a timer value of the timer. SCSI running when the timer times out
A device control apparatus, comprising: an abort processing unit that aborts a command without resetting the target. 4. Setting means for setting the type of event that needs to be processed and inspection means for inspecting the occurrence of the event are provided, and only when the event has occurred when the timer has timed out, The device control apparatus according to claim 3, wherein the abort processing unit is caused to perform the abort processing. 5. An inspection means for investigating the presence or absence of a failure of the device is provided, and after the abort processing means performs the abort processing, the investigation means conducts the inspection of the device. The device control device according to 1.