JPS60245029A - Data write system - Google Patents

Abstract

PURPOSE:To improve an access time which is viewed externally, to prevent write data from being lost owing to abnormality of hardware, and to improve its system efficiency by providing a nonvolatile memory for storing the write data temporarily. CONSTITUTION:When data is written in a disk unit 4, write data transferred from a host CPU through a host path 7 is stored in the nonvolatile memory 6 under the control of a local CPU2 and a disk controller 3. The CPU2 accesses the address of a cylinder head and an address where the write address is written in its internal memory. Then, the distance between both addresses is compared with a specific value and whether the unit is in a writable state or not is checked when the distance is larger than the specific value; when so, the data stored in the memory 6 is written in the target cylinder. Consequently, the access time viewed from the host CPU is shortened, the write data is prevented from being lost, and the system efficiency is improved.

Description

DETAILED DESCRIPTION OF THE INVENTION [Technical Field] The present invention relates to a data writing method in a disk system.

r Prior Art] In conventional disk systems, data is written to the disk using one of the following methods.

■ When receiving a seek command for data writing, the machine immediately seeks to the target cylinder and writes data, regardless of the head position at that time.

(2) Write data is temporarily stored in a volatile memory that also serves as a buffer memory for read data, and the write data is written to the disk during free time during data transfer with the host CPU. The seek operation is performed regardless of the head position, similar to the method (2) above.

The method (2) described above has the disadvantage that the seek time may be long and the access time to the disk viewed from the outside cannot be shortened.

In the method (2) above, the access time to the disk viewed from the outside is shortened. However, since the write data is stored in a volatile memory, there is a risk that the write data may be lost before it is written to the disk due to a momentary power outage, hardware abnormality, or the like. In addition, since writing data stored in volatile memory to the disk is performed in the same way as method ``■'' above, when viewed from within the system, there is a problem of increased seek time similar to method ``■'' above, resulting in system efficiency. cannot be improved).

[Purpose] The purpose of the present invention is to provide a data writing method in a disk system that can reduce access time seen from the outside, prevent loss of written data due to hardware abnormality, and improve system efficiency. .

〔Example〕

An embodiment of the present invention will be described below with reference to one drawing.

Embodiment 1 FIG. 1 is a schematic block diagram of a disk system. l
is the interface part with the host CPU.

2 is the local CPU, 3 is the disk controller.

4 is a disk unit, 5 is a volatile memory, and 6 is a nonvolatile memory. 7 is a host bus between the host CPU and the host CPU.

Commands given from the host CPU through the host bus 7 are sent to the local CPU via the interface section 1.
j2, where it is interpreted, converted into another command format, and sent to the disk controller 3. Disk controller 3.

The disk unit 4 is controlled according to the command.

If the command is a read command, a seek to the target cylinder is performed under the control of the disk controller 3, the read gate signal of the disk unit 4 is turned on at an appropriate timing, and successive data is transferred to the disk controller 3.
The data is stored in volatile memory 5 via. The read data stored here is transferred to the host CPU via the interface section 1 under the control of the local CPU 2 in synchronization with a DMA request from the host CPU. Next, when there is a request from the host CPU.

The local CPU 2 checks whether the data is stored in the volatile memory 5, and if so.

The data is transferred to the host CPU without accessing the disk.

Writing data will be explained. Write data transferred from the host CPU is temporarily stored in the nonvolatile memory 6 under the control of the local CPU 2 and disk controller 3. Writing of the write data stored here onto the disk and seeking to the target cylinder is executed when the head approaches within a certain distance of the target cylinder, as shown in the flowchart of FIG.

That is, the local CPU 2 calls the cylinder address A where the head is located and the cylinder address B where data is to be written from the internal memory during a free time for data transfer with the host CPU (step 10.11). The distance between cylinder address A and cylinder address B is compared with a predetermined value n (step 12), and if the distance is greater than or equal to n, it is checked whether the disk unit 4 is in a writable state (step 13). If writable, - a command is issued from the local CPU 2 to the disk controller 3 to seek and write data to the target cylinder;
Under the control of the disk controller 3, one write data stored in the nonvolatile memory 6 is written to the target cylinder (step 14). Thereafter, it is checked whether all the write data in the nonvolatile memory 6 has been written (step 15), and if data remains, the process from step 10 is repeated. Similarly, in step 13, if writing is not possible, the process returns to step IO.

In this way, the write data from the host CPU is immediately temporarily stored in the nonvolatile memory 6, so that the access time seen from the host CPU is sufficiently shortened. Also, there is a write operation of write data once stored in the non-volatile memory 6 to the disk.

Thereafter, the process is executed when the head approaches within a certain distance of the target cylinder (due to a previous read operation, etc.), so it is possible to suppress the seek time within a certain range and improve system efficiency. Furthermore, the write data stored in the nonvolatile memory 6 is preserved even if a power supply or hardware abnormality occurs.

Example 2 Figure 3 is a schematic block diagram of the disk system.

1, and the same parts as in FIG. 1 are given the same reference numerals. In this embodiment, as shown in FIG. 4, a plurality of blank cylinders 9A, 9B, 9C,
9D and use them as non-volatile memory to temporarily store write data.

Data reading is the same as in the first embodiment, so its explanation will be omitted, and data writing will be explained with reference to the flowchart of FIG. 5.

Commands from the host CPU are interpreted by the local CPU 2, converted into command formats, and sent to the disk controller 3. When the local CPU 2 receives the seek command, the local CPU 2 reads the cylinder address A where the head is located and each blank cylinder address from the internal memory (step 20.21), and reads the blank cylinder closest to the cylinder address A. Select address B (step 23). And the distance IA between cylinder addresses A and B
- Calculate Bl (step 23). Similarly, the distance IA between the target cylinder address C specified by the command and the cylinder address A is called from the internal memory.
-Cl is calculated (step 25), and the two distances are compared (step 26).

If IA-Bl≧lA-CI, the local CPU 2 instructs the disk controller 3 to seek to the target cylinder address C (step 27).

After the seek is completed, the write data is written to the designated sector of the target cylinder.

If IA-Bl<IA-CI, the local CPU 2 instructs the disk controller 3 to seek to blank cylinder address B (step 28).

After the seek is completed, the write data is written to the blank cylinder.

For example, in FIG. 4, when the head 17 is in the position shown, if a seek command to the target cylinder 16 is received, it will seek to the blank cylinder 9B, and after the seek is completed, the write data will be stored in the blank cylinder 9B. become.

The write data stored in the blank cylinder is then written to the designated sector of the target cylinder under the control of the local CPU 2. That is, when the head approaches within a certain distance of a blank cylinder in which loaded data is stored during idle time for data transfer with the host CPU, the local CPU 2 instructs the disk controller 3 to transfer data to that blank cylinder. The write data stored therein is read out to the small capacity buffer memory 18 in the disk controller 3. Next, when the head approaches within a certain distance of the target cylinder for the write data, it issues an instruction to the disk controller 3 to seek to the target cylinder, and transfers the write data in the small capacity designated buffer memory to the designated target cylinder. Write to sector.

In this embodiment as well, by providing blank cylinders at appropriate intervals, the seek time can be kept within a certain range and the access time seen from the outside can be shortened.

It can also improve system efficiency. It is also possible to prevent write data from disappearing due to power failure or hardware failure. moreover,
Since a specific cylinder on the disk is used as a non-volatile memory for temporarily storing write data, the cost can be lowered than when a separate non-volatile memory is provided.

Note that it is also possible to write the write data stored in the blank cylinder into the target cylinder without considering the head position. However, it cannot be denied that the system efficiency is lower than in the above case due to the increase in seek time for reading the blank cylinder and writing to the target cylinder.

〔effect〕

As described in detail above, according to the present invention, it is possible to shorten the access time of a disk system when viewed from the outside, improve system efficiency, and prevent write data from being lost due to hardware abnormalities or the like.

[Brief explanation of drawings]

FIG. 1 is a schematic block diagram of a disk system according to an embodiment of the invention, and FIG. 2 is a flowchart for explaining a data write operation of the embodiment. Fig. 3 is a schematic block diagram of a disk system according to another embodiment of the present invention, Fig. 4 is a conceptual diagram for explaining blank cylinders and head positions, etc., and Fig. 5 is a data writing in the same embodiment. It is a flowchart for explaining operation. l...Interface unit, 2...Local memory, 3...Disk controller, 4...Disk unit, 5...Volatile memory, 6...Non-volatile memory, 7...Host bus , 9A-9D...Blank cylinder, 16...Target cylinder, 17...Head. Figure 5

Claims (3)

[Claims] (1) In a disk system, a non-volatile memory for temporarily storing write data is provided, and the write data is temporarily stored in the non-volatile memory. A data writing method characterized in that the written data is then written to a disk. (2) Claims characterized in that when the head approaches within a certain distance a target cylinder for write data stored in a non-volatile memory, the head seeks to the cylinder and writes the write data there. Data writing method described in Section 1. (3) Multiple blank cylinders are provided on the disk as nonvolatile memory for temporarily storing write data, and when a seek command for data writing is received, the head and the blank cylinder closest to it are connected. Compare the distance with the distance between the head and the target cylinder specified by the seek command, and if the former distance is greater than or equal to the latter distance, immediately seek to the target cylinder and write the write data, otherwise, seek to the target cylinder and write the write data. Seek to and write the write data there once, then,
3. The data writing method according to claim 1 or 2, characterized in that data is written to the target cylinder.