JPH0784899A - Selective resetting system for scsi equipment - Google Patents

Abstract

PURPOSE:To reset only a SCSI equipment, selected from among plural SCSI equipments connected to a SCSI bus, by using a forcible reset instruction such as a 'RESETCONDITION' instruction. CONSTITUTION:The respective SCSI equipments are provided with a selector which selects one data line of a SCSI data bus according to the device ID number of the corresponding equipment and the equipment reset by ANDing the output of the selector with the forcible reset instruction from the reset signal line of the SCSI bus. As a 2nd system, the respective SCSI equipments are provided with a means which confirms a host that is connected and only the SCSI equipment whose connection with the host having sent out the forcible reset instruction can be confirmed is reset with the forcible reset instruction.

Description

Detailed Description of the Invention

[0001]

[Industrial application] One or more personal computers and hard disks, magneto-optical disks, CD-
SCSI (Small Computer System Interfac) is a device connection method for small computer systems that can connect up to eight peripheral devices such as ROM drives in total.
e) Connection methods are commonly used.

The present invention uses the forced reset command used in such a SCSI connection system to enable the SCSI
The present invention relates to a selective reset method for resetting only a specific SCSI device selected from a plurality of SCSI devices connected to a bus.

[0003]

2. Description of the Related Art FIG. 5 conceptually shows a SCSI connection system, and shows a state in which eight SCSI devices # 0 to # 7, which are the maximum number of SCSI device connections, are connected by a SCSI bus SB. , These SCSI devices each have a SCSI interface part IF.
CSI devices # 0 to # 7 are interface parts
It is connected like a potato by the SCSI bus SB via the IF. The end of this SCSI bus is terminated by a terminating device (not shown).

FIG. 6 shows each SCSI as described above.
2 is a diagram showing a configuration of a portion related to the present invention of an interface unit IF provided in a device, which is a driver / receiver for inputting / outputting data to / from a host device such as a computer or a lower device such as a hard disk. And a microprocessor MPU that controls according to a program based on the SCSI protocol stored in ROM, a SCSI protocol controller SPC that controls the SCSI bus SB according to the control of this microprocessor MPU, and a control of the above microprocessor MPU. Therefore, this driver / receiver and this SCSI
It has a data transfer circuit that transfers data between protocol controllers SPC.

[0005]

The reset process in this SCSI connection system will be described below.
When a failure occurs in an I-device (hereinafter referred to as a device), the device that detected this failure is called "BUS DEVICE RES".
S to send a reset instruction message such as "ET" or "ABORT"
Send it to the device where the fault is found via the CSI bus,
The device that receives this message is configured to determine the ID of the host that sent the message and reset the command from this host.

However, if the failure of this device is an interface time-out failure caused by the hang-up of the MPU of the interface section, etc., the MPU itself cannot respond to the above reset instruction message. The device reset cannot be recovered by the reset instruction message.

In such a case, a device such as a host that has found a failure generates a "RESET CONDITION" command and sends it to the SCSI bus, while the "RESET CONDITION" command is generated.
Resetting all SCSI devices that receive a command is defined as a standard of the SCSI connection system.

Therefore, since all the SCSI devices connected to this SCSI bus are initialized, all the processes being executed by these SCSI devices are interrupted.

As shown in FIG. 7, if the # 2 device personal computer serves as a host, a format command is sent to the # 5 device hard disk.
When this hard disk is formatting the SCSI bus SB while disconnecting, # 1
It is assumed that, as a result of another personal computer, which is a device, accessing as a host for reading from the magneto-optical disk, which is a device # 5, an unrecoverable failure is detected in the device by the reset instruction message as described above.

Then, as described above, from the personal computer which is the # 1 device, "RESET CONDITION"
The command is sent to the SCSI bus SB and the SC
There is a problem that all devices connected to this SCSI bus SB are reset, including the hard disk (# 6 device) that is being formatted in the disconnection state that is temporarily disconnected from the SI bus.

Therefore, in order to complete the formatting of the hard disk which is the # 6 device, this "RESE
After all devices have been reset by the "T CONDITION" command and this system has been restored, the format command is sent again from the personal computer of the # 2 device, which is the host of this # 6 device, to this # 6 device. It is necessary to restart the process from the beginning.

In order to solve the above-mentioned problems, the present invention solves all the problems that all SCSI devices connected to a SCSI bus are connected even when a plurality of SCSI devices are connected to the SCSI bus.
An object of the present invention is to make it possible to selectively reset only required devices by using a forced reset command (corresponding to the above-mentioned "RESET CONDITION" command) for forcibly resetting the CSI device.

[0013]

In a SCSI connection system including a plurality of SCSI devices each connected to a SCSI bus via a SCSI interface unit, each SCSI device has one of the data lines of a SCSI data bus. Is provided in accordance with the device ID number of the device, and an AND circuit for taking the logical product of the output of this selector and the forced reset command from the reset signal line of the SCSI bus is provided, and the output of this AND circuit is SCS.
It is supplied as a reset signal of the I protocol controller.

In addition, as a second method, each is SC
In the SCSI connection system including a plurality of SCSI devices connected to the SCSI bus via the SI interface unit, a means for confirming the host connected to each SCSI device is provided, and the device connected to the host that has issued the forced reset command is connected. Only the SCSI device that has been confirmed to be able to perform the reset operation by the forced reset command can be performed.

Further, in the second method, the SCSI device serving as the host is provided with a counting means for counting the number of the forced reset commands repeatedly sent out due to the failure of the target, and the count value of the counting means is predetermined. When the value exceeds the value, the use of the target in which the failure has occurred may be prohibited.

[0016]

Each of the inventions described in claims 1 to 3 of the claims of the present application has a plurality of SCSs, each of which is connected to a SCSI bus through a SCSI interface unit.
In the SCSI connection system equipped with I-device, "RESET CO
It is configured such that only the selected SCSI device can be reset by using a forced reset command such as the "NDITION" command.

According to the first aspect of the invention, SCSI is used.
Each of the data lines of the bus is assigned according to the identification number (ID) of each SCSI device, and the selector provided in each SCSI device selectively outputs the signal on the data line according to the identification number of the SCSI device. The AND circuit that logically ANDs the output of the selector and the forced reset command from the reset signal line of the SCSI bus outputs the output from the selector and the forced reset command at the same time to the reset signal of the SCSI device. Supply as.

Therefore, even if a forced reset command is issued to the reset signal line of the SCSI bus, the SCSI device not selected by the data line is not reset, so that the processing being executed is not interrupted or canceled. .

According to the second aspect of the present invention, when each SCSI device receives the forced reset command, it is confirmed whether or not there is a SCSI device as a connected host, and if it is connected to the host. If it is confirmed that the reset operation is performed by this forced reset command.

That is, a certain SCS by a certain host
The fact that the I-device has been found to be faulty is because this SCSI device is in a state of being connected to the host that has found the fault, and therefore the SCSI device in which the fault has been found is in a connection state with this host. It will be apparent that the fault can be recovered by resetting only this SCSI device by the forced reset command from this host.

When a plurality of hosts such as a plurality of personal computers are connected to the SCSI bus,
Generally, since the SCSI device stores the identification number of the connected counterpart device, it is necessary to perform the reset operation only when the identification number of the host that issued the forced reset command matches the stored identification number of the counterpart device. The required recovery processing has been performed.

As described above, also in the second aspect of the invention, the SCSI device which is performing the processing such as formatting in the disconnected state disconnected from the SCSI bus can continue the processing being executed, and after resetting. There is no need to re-process.

Further, in the invention according to claim 3, in the invention according to claim 2, when the SCSI device in which the failure has occurred cannot be reset even by the forced reset command, the forced reset command is repeated many times. Since the SCSI bus is sent from the host, it is prevented that the SCSI bus is substantially occupied by the forced reset command and cannot be processed by another SCSI device.

That is, according to the third aspect of the invention, when the number of issuance of the forced reset command from the host exceeds a predetermined number, the use of the faulty SCSI device is prohibited and the forced reset command is issued. It is configured to stop the sending and restore the system function.

[0025]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 1 shows a first embodiment of the present invention in which the hardware of each SCSI device can achieve the object of the present invention. Reset I
The structure of a D identification circuit is shown.

The reset ID identification circuit has input terminals connected to the data lines DT0 to DT7 NI of the SCSI data bus, and a selector for selecting the input terminal according to the device ID input to the selection control terminal. It is composed of an AND circuit which takes the logical product of the output of the selector and the forced reset command from the reset line of the SCSI bus.

If the reset ID identification circuit is the device I,
If D is “3”, it is assumed that it is provided in the interface section of the SCSI device of # 3, and when a “1” level signal is applied to the data line DT3 of the SCSI data bus, a “1” level signal is output from the selector. A signal is output, at this time,
If a forced reset command is applied as a "1" level signal to the reset line of the SCSI bus, an "1" level output is generated from the AND circuit. Therefore, the output of this AND circuit is applied as a reset signal to the SCSI protocol controller. This resets the SCSI protocol controller.

At this time, in the reset ID identification circuit provided in the SCSI device other than # 3, since the device ID of the SCSI device is a number other than "3", the data line DT3 of the SCSI data bus. Even if a "1" level signal is applied to the selector, the output of the selector is "0", and therefore the "1" output from the AND circuit does not occur. Therefore, the reset signal is sent to the SCSI protocol controller of the SCSI device other than # 3. Will never be supplied.

Therefore, as described above with reference to FIG. 7, the hard disk of the # 5 device is the SCSI bus SB.
However, while formatting is being executed in the disconnected state, the hard disk formatting process will continue even if the "RESET CONDITION" command, which is a forced reset command, is sent from the # 1 device personal computer. It is not necessary to send the format command again and restart the formatting from the beginning.

Data corresponding to the device ID, “3” in the above example, is, for example, the MPU of the interface unit (FIG. 6).
Needless to say, it can be taken out from the internal register or the device ID setting switch provided in the SCSI device.

FIG. 2 is a flow chart showing the processing using a host having the # 3 device as a control target (target) as an example. As shown in step [11], the SCSI device connected to the SCSI bus at the time of start-up. Command, issue a command to the target as shown in step [12] as a normal process, and determine whether the process of this command ended normally in step [13]. Issue the following command:

However, if the command processing is not completed normally, the process moves to step [14] to issue a sense command to the target, and in step [15] it is judged whether or not this target hangs up. If it is not hung up, it is judged in step [16] whether or not this target should be reset. However, depending on the reason why it does not end normally, it returns to step [12] and issues a command.

When it is determined in step [15] that the target is hung up, or when the target is reset according to the determination result in step [16], the present invention proceeds to step [17]. Data line D of SCSI data bus as shown
It sends a "1" level select signal to the data line DT3 corresponding to the target of T0 to DT7, and sends a forced reset command of "1" to the reset line of the SCSI bus.

Step [18] is a reference to the operation of the target. Since both the selection signal and the forced reset command are at "1" level, the AND circuit outputs "1" level as described with reference to FIG. Is supplied to the SCSI protocol controller, the target of # 3 is reset and recovery processing is performed.

In FIGS. 3 and 4, all the SCSI devices are not reset just by the arrival of the forced reset command.
FIG. 3 shows a second embodiment of the present invention in which only the SCSI device that has been confirmed to be connected to the host that has sent this forced reset command is reset. FIG. 4 is a flowchart of the processing in the target.

When the host detects the occurrence of a failure in the target (step [21]), it sends a reset instruction message such as "BUS DEVICE RESET" or "ABORT" to the target in step [22] as in the conventional case. If it is determined in step [23] that the reset is successful, the process proceeds to step [24] to perform error processing similar to the conventional one and recover.

However, if it is determined in step [23] that the reset by the reset instruction message is not successful, for example, the above-mentioned "RESET CONDITION"
Instructs the SCSI protocol controller SPC to send a forced reset command corresponding to the command to the SCSI bus, and as shown in step [26], the identification number (ID) of the faulty device is controlled by the MPU of the interface unit. Store in register etc.

In the next step [27], the SCSI protocol controller SPC is instructed to send a selection signal for selecting the device having the identification number stored in the above step [26], and after this selection is completed, the step [2
In 8], send a forced reset command to the SCSI bus.

At the target, as shown in the flow chart of FIG. 4, when the selection signal is received at step [42] and the forced reset command is received at the next step [43], the connection is continued at step [44]. Check if there is another device (host).

If it is determined in this step [44] that there is a host that is connected, recovery is performed by resetting the process being executed by a command from this host in step [45]. If the device is not connected or the connection cannot be confirmed during the selection process, the process proceeds from step [44] to step [46] to continue the process being executed.

As described above, if the target is not reset by the forced reset command, the host obtains the result of unsuccessful reset at step [29] of FIG. After adding "1" to the count value N of the counter that counts the number of successes,
In the next step [31], it is determined whether or not this count value (N + 1) exceeds a predetermined value K.

As a result, if (N + 1) .ltoreq.K, the number of reset failures is less than a predetermined value. Therefore, returning from step [31] to step [25],
The processing from step [25] to step [31] is repeated until the result of step [31] becomes (N + 1)> K.

If the reset of the target is successful during this repeating period, the process moves from step [29] to step [33] to perform the existing error processing, and at the next step [34], the above step [26] is performed. The ID of the target in which the failure has occurred, which was stored in step 5, is erased and the normal processing is returned to.

However, if the reset is not successful even if the number of repetitions from step [25] to step [33] exceeds the predetermined number K, the device in which the failure has occurred is determined in step [32]. Thereafter, the use is prohibited and the process returns to the normal process.

The steps [47] and [48] in FIG.
This corresponds to the normal reset instruction command sent in step [22] of the host in FIG. 3, and if the target is reset by these steps, the process moves from step [23] of the host to step [24] Perform the error processing of and then recover.

[0046]

According to the present invention, the forced reset instruction is S
Even if it is sent to the CSI bus, the required SCSI device is only reset and other SCSI devices are not reset, so there is no need to reset or restart a normal SCSI device. Not only that, the processing that was being executed when this forced reset command was sent is also continuously executed until this processing ends, so a special effect that the processing efficiency of the system is dramatically increased is achieved. To be done.

[Brief description of drawings]

FIG. 1 is a block diagram showing a first embodiment of the present invention.

FIG. 2 is a flowchart of processing of a host according to the first embodiment of this invention.

FIG. 3 is a flowchart of a host in the second embodiment of the present invention.

FIG. 4 is a flowchart of a target in the second embodiment of the present invention.

FIG. 5 is a conceptual diagram of a SCSI connection method.

FIG. 6 is a schematic diagram of a SCSI interface unit.

FIG. 7 is an operation example of a SCSI connection method.

Claims (3)

[Claims] 1. A SCSI connection system comprising a plurality of SCSI devices, each of which is connected to a SCSI bus via a SCSI interface unit, wherein each SCSI device has one of the data lines of a SCSI data bus as a device of the device. A selector that selects according to the ID number and an AND circuit that ANDs the output of this selector and the forced reset command from the reset signal line of the SCSI bus are provided, and the output of this AND circuit is used as the reset signal of the SCSI protocol controller. A selective reset method for SCSI devices characterized by being supplied. 2. A SCSI connection method comprising a plurality of SCSI devices, each of which is connected to a SCSI bus via a SCSI interface unit, wherein each SCSI device has means for confirming the host to which it is connected, and a forced reset is performed. A selective reset method for a SCSI device characterized in that only the SCSI device that has been confirmed to be connected to the host that has issued the command performs the reset operation by this forced reset command. 3. The SCSI device serving as a host is provided with counting means for counting the number of forced reset instructions repeatedly sent out due to a failure of the target, and when the counted value of this counting means exceeds a predetermined value. The selective reset method for a SCSI device according to claim 2, wherein use of the target in which the failure has occurred is prohibited.