JPS58205269A - Controlling method of rotation type memory - Google Patents

Abstract

PURPOSE:To evade the transfer omission of a time-series signal by eliminating the need to take a long time for access to a substitute sector even when a defective spot characteristic to a rotation type memory is present on a storage medium. CONSTITUTION:When a track write command is supplied to an input port 102 externally, a CPU 114 selects a specified track in response to the command. If the write command is supplied while a head runs on some sector on some track, the CPU 114 detects the next index to the sector to discriminate the address of the new sector. A substitute table is searched by the sector address and when the sector is an unused sector 412 or a bad sector, the next sector is waited for. When the sector is usable, a status display part is checked and when a bit W is ''0'', data is read out of a buffer 106A or 106B and written in the sector.

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a memory control method, and particularly to a rotary memory control method for controlling a rotary memory such as a magnetic disk, an optical disk, or a magnetic drum. Magnetic disks include disk packs, data monols, floppy disks, and the like.

These rotation+(lq) memories have so-called bad spots, where information cannot be recorded correctly in a part of the storage medium due to, for example, a manufacturing defect.

It often includes For example, when writing data to a magnetic disk, if there is a sector (bad sector) containing such pads, a spare sector of an alternative track is used. Generally, this alternative track is often assigned to the outermost track of the disk. When a bad sector is identified while writing data to a certain track, the magnetic head is moved to the outermost track, an alternative sector is selected, data is written, and the magnetic head is moved to the original track. The same applies when reading data. For example, if the average access time on one track is 10 ms, and the time required for a magnetic field to move between the outermost track and the original track is 50 ms each way, then each bad sector will have 110 ms. It will cost you milliseconds of extra time. This is comparable to the time required to transfer two troughs of data, approximately 100 milliJ seconds.

In a typical processing system, control of a magnetic disk is performed by a data channel independently of the central processing unit (CPU). The memory control section of the channel is C
Failed to read channel commands given from PU sequentially,
The overall processing speed is improved by reading data all at once from sectors that can be accessed with the least amount of access time, or by storing data therein. Since processing in a normal CPU generally allows for queuing, there is no problem even if it takes a particularly long time to access the magnetic disk due to the existence of bad spots.

However, in systems where a large amount of time-series signals must be input/output to/from rotating memory while maintaining a predetermined data transfer rate, delays in memory write/read operations due to the presence of bad spots can hinder data transfer. will give.

For example, in an image data processing system for medical diagnosis, a radiation image recorded on a phosphor material with so-called photostimulability is scanned with a laser beam to obtain stimulable luminescence, which is then photoelectrically converted into an input image. Image processing is performed on the series signal to obtain an output time series I1, and the signal running on the recording medium is modulated by the output time series I1 to produce a visible image p). In this system, for example, 4 megabytes per frame of a radiation image is determined by which pixel y'-data is used, and the pixel data of one frame or multiple frames before and after is temporarily stored, and pixel data between multiple frames is stored. For example, a magnetic disk is used as a temporary storage medium because of its large amount of information for the purpose of image processing or obtaining a reproduced image multiple times by changing the image processing conditions of one frame.

In such image data processing systems, it is often necessary to maintain the data transfer rate of input time series signals and output time series signals constant due to optical or mechanical limitations of the image reading device and image recording device. . Normally, these image reading devices and recording devices are unable to interrupt the recording operation in the middle of the process once they start processing, since it is necessary to maintain the quality of the image. For example, in a system that uses optical scanning for the main scanning of images and mechanical scanning for the sub-scanning, there are some problems with the mechanical operation of the device, which is not economical with current technology. As long as components that can be realized in In a system in which, for example, an alternative sector of the outermost track must be accessed every time a bad spot is detected, there is a risk that data will be lost due to the long access time.

The present invention eliminates the drawbacks of the prior art and allows time-series input/output signals to be processed at a predetermined transfer rate without causing deletions in the time-series input/output signals even if the storage medium has a defective location that prevents normal storage. The purpose of this invention is to provide a control method for a rotating memory that can transfer data.

This object is achieved by a rotating memory control method embedded in the present invention as follows. That is, this is a control method for a rotary memory that reads data from and stores data 11'W- in a memory 11'W- of a rotary memory.
; When preparing a first table indicating the existence or nonexistence of a storage section that cannot be stored normally in the storage unit corresponding to C billion units, and reading or writing data to the storage position jIi, Refers to the first table and, if there is a storage section in the storage unit that cannot be stored normally, skips that storage section and reads or writes data to the next storage section in the same storage unit where it can be stored normally. It is.

Note that in this specification H, a "storage unit" is a control unit of a storage location accessed in a rotating memory, and may include one or more tracks in a magnetic disk, for example.

Furthermore, the term "storage section" refers to the minimum access unit of data constituting a storage unit, for example, a sector on a magnetic disk.

Next, with reference to the accompanying drawings, one example of a rotary memory according to the present invention will be described.
I will explain the DIJ method in detail separately.

FIG. 1 is a schematic diagram conceptually showing the basic configuration of a simultaneous transfer type memory system (incorporated) device according to the present invention.

This device has a memory control section 100, and a memory control section 10.
0 is connected to an input port 102 for receiving a time-series input signal, such as, for example, an image rli 1m, and an output port 104 for transmitting such a signal as a time-series output signal. The memory controller 100 is also connected to a rotating memory 200, such as a magnetic disk, via a buffer memory 1,1106. -For example,
Rotating Memo', +200 takes 20 minutes to sell once
The zero buffer memory 106, which is a magnetic disk device with an adjacent track moving time of 10 milliseconds, is composed of two banks 106A and 106B each having a data storage capacity of one track of the magnetic disk 200.

The memory ili control unit 100 has a function of controlling reading/writing of the magnetic disk 200 while maintaining a transfer rate of transferring two L troughs of data as a time-series signal in, for example, 100,917 seconds.

FIG. 2 shows a specific configuration of one embodiment of the rotary memory limiting device shown in FIG. 1. In FIG. In this figure, components similar to those in FIG. 1 are designated by the same reference numerals.

The memory control unit 100 includes a time series signal input/output capacitor) 102
and an input/output circuit 110 connected to the central processing unit (C
PU) 114. Bus 112 symbolically represents a data bus, address bus, command bus, and other control lines to avoid complicating the diagram. The bus 112 includes a read-only memory (ROM) that stores 1ltlJ control programs for the CPU 114, fixed r-control programs, etc.
) 116 '1 Random access memory (RAM) 118 as high-speed memory, and the buffer memo of FIG. J106 is connected. In the figure, the paranoia memory 106 is connected to the outside of the memory control suppressor 100.
8'&q', but if the memory control unit 100 (as shown in FIG. 2) is realized with a processing device such as a microcombi, RAM
Part of the area within 118 or the same t1 as RAM 118
Since it can be implemented using a separate RAM, it can be considered as part of the memory 1li111111 section 100 as shown in FIG.

The magnetic disk device serving as the rotating memory 200 has a second
In the figure, a disk unit 202 such as a disk pack, a disk circuit 204 including a magnetic head control circuit Σ-17 data writing/reading circuit, and data lines, address lines, control lines, etc. of the memory control unit 100 are connected. The interface unit 206 has an interface section 206 for controlling the interface. The interface section 206 is connected to the bus 112 of the memory control section 100 and can access the paranoia memory 106 via the data line 120.

In addition, in this embodiment, the li control section 100 shows ``NO'' in one microcopy 7 pins, but this
・It goes without saying that a single 12-microcomputer of the strip-and-chair type may be used, and that it can also be realized by a wired logic circuit instead of a microconbeater.

According to the present invention, a bet sector table 300 and a replacement table 400 are prepared in the storage area of the RAM 118 (FIG. 3). The bet sector table 300 shows the track number and sector signal of the bet sector.
An example is shown in the figure. In this example, disk pack 20
2 is two units 202-0 and 202-1 (third
As shown in Figure 2, each unit uses six disks as an effective storage area. As shown in FIG. 4, if 6 bits are assigned to the sector signal display section 302 and 10 bits are assigned to the track number display section, and one word is used for 12 sectors, the disc 1. Assuming that the maximum number of pad sectors per surface is 20, the pad sector table 300 for 2 unit, 202-0 and 202-1 is 24.
Use the high-speed Nomori pirate with 0 words (Figure 3).

Busk units 202-0 and 202-1 typically have their suppliers pre-detect the presence of a pad spot and often pre-record its location on a particular track 210. Of course, the pad spot of the disk may be detected at the time of initialization in the usage sound III, and information regarding the pad sector may be written in a specific track 210 of the disk unit. track 2
The information regarding the last sector recorded in 10 is changed to 9 when the disk initialize operation or the system power is turned on.
, the CPU 114 activates a subroutine for setting a sector table, so that the RAM is transferred from the track 210.
118 and is stored in the area 300. This is indicated by arrows 212-0 and 212-1 in FIG.

As shown in FIG. 4, in the bet sector table 300, the position of the bet sector is displayed in the word corresponding to the unit number and head signal as the smell of the bet sector and the 7th position of the track.

If there is no bad sector, all r - I J (3F in 6-bit hexadecimal, 3FF in to-bit hexadecimal)
) will be left alone. In the example of FIG. 4, bet sectors are displayed at two locations 310 and 312.

The replacement table 400 is a table that establishes the correspondence between logical sectors and physical sectors and eliminates the use of solid sectors. This determines which track will be used to read or write data, and while seeking that track,
or in the previous free time, that is, each time the track is updated, the RAM is
118 area 400. This is indicated by arrow 214 in FIG.

According to the present invention, an alternative sector for a bet sector is provided in a track included in the bet sector. Typically, 1) a block operation is performed in one sector so that, for example, at least two sectors remain unused. As shown in an example in FIG. 5, the replacement table 400 includes a 4-bit number display section 402, a 6-bit sector number display section 404, and a status display section 406. A word corresponds to one sector. In this example, one track has RAM
118, an area 400 of 50 words is allocated (FIG. 3). All l1 for old sectors and unused sectors
1 (3F in 6-bit hexadecimal notation) is stored. Fifth
In the example shown in the figure, physical sector 3, that is, the sector at the position indicated by 410, is a solid sector, and thereafter, physical sector 4 becomes logical sector 3, physical sector 5 becomes logical sector 4, and so on. The correspondence between logical sectors and physical sectors is shifted so that some of the sectors are used as alternative sectors. The sector indicated by the number 412 is an unused sector. In addition, as shown in FIG.
AW, and each of these hits indicates whether or not read, write, and read-after-write operations have been completed for each sector. These beats are "1"
indicates the end of the operation, and "0" indicates unfinished.

As mentioned above, when initializing or turning on the power to this device, the pad sector table 300 is set to RA.
Created in M 118.

For example, in the case of a search operation, once the track of the magnetic disk 202 to be accessed is determined, the CPU 114 moves the magnetic head to that track and replaces that track while the seek operation is being performed or in the free time just before that. A table 400 is created in RAM 118. The time series signal inputted to the input Z-tooth 102 at a predetermined transfer rate is
is accumulated in The next one track is the other puncture 106
B is sequentially stored, but in the meantime, the data for one track that has been stored in the back 106A is punctured 106.
The data is read from A, transferred to the magnetic disk device 200 via the data gland 120, and transferred to the selected portion of the magnetic disk 202 through the interface section 20G and the disk circuit 204. . At that time, Cr'U
114 refers to the replacement table 40G, and if there is a pad sector, it is skipped, and data is written to the next available sector in j order. ■ Punk 1 track worth of data
When 06A finishes transferring to that track, CPU 11
4 selects another track and performs the same operation.

- Detailing a write operation as an example. When a track write command is given to the input port 102 from the outside, C
In response, PU 114 selects the designated track. If a write command is given while the head is running on a certain sector of a certain track, the CPU 11
4 detects the next index of that sector to identify the address of that sector, ie, a new sector.

The replacement table 400 is searched using this sector address, and if the sector is an unused sector 412 or a bad sector 410, the next sector is waited for. If the sector is available, the status display section 40
6, and if the W bit is "O", data is extracted from Balanor 106A or 106B and written into that sector, and the W bit is changed to "1". RAW bit is “
If it is "1", that sector is a sector in which writing has been completed normally, so the process waits until the next sector. If the RAW bit is rOJ and the W focus is "1", this is a sector in which only a write operation has been completed, so the contents of the sector are read out and checked for errors. That is, if an error is detected, the W pit in that sector is set to "o" and the sector is used for the next write operation. If the department is not detected, RA
When the W bit is set to "1", it means that the sector has been damaged and the writing has been completed normally.

In this way, when a write command is given, even in the middle of a track, the write operation can be performed immediately from the next arriving sector without waiting for the return of the leading sector of the track. Also, since the prime sector of sector 1 and 1 is on the same track, only one seek operation is performed per tiger. Therefore, in the above example, 1
1. Since 10 milliseconds are imprinted on the transfer time of the program and the seek operation between adjacent tracks is 10 milliseconds, the difference of 90 milliseconds is allocated to one track. designated and accessible times. During this time, the disk rotates 45 times, so write and read-after-write can each be attempted twice.

In the case of a read operation, once the track of the disk 202 to be accessed is determined, the replacement table for that track is stored in the RAM.
118. Next, data is read out from that track of the guide 202 through the disk circuit 204 and the ink face section 206 2,
This is stored in one puncture, for example 106A, through 0. At that time, CPU t 14 refers to the exchange table 400 of that track, and if the bad security wants to skip it, Il, ii "2nd" Read the data of all 1 history sectors in the track. ■ When the track is finished storing and transferring the portion of the data to the punk 106A, the CPU
114 selects another '◇ track and performs the same operation. While data is being written from one track to the other bank 106B, the data previously stored in one bank 106A is read out from the bank 106 and sent out from the output port 104 at a predetermined transfer rate. .

The rotating memory control method according to the present invention does not require time to access an alternative sector even if bad spots, which are unique to rotating memories, exist in the storage medium. can be read/written to a rotating memory while maintaining a predetermined transfer rate.

Therefore, no gaps occur in the transfer of time-series beliefs.

The invention finds particular application in data processing systems in which a constant data rate is output.

[Brief explanation of the drawing]

Fig. 1 is a schematic diagram conceptually showing the no-two-piece configuration of the rotary memory control device according to the present invention, and Fig. 2 shows a specific implementation in which the device shown in Fig. 1 is realized by a microcombi. Block Diagrams Showing Examples FIGS. 3 to 6 are configuration diagrams showing the configurations of a bad sector table and a replacement table. Explanation of one code of the main parts ioo...Memory discount 1 section 106...Buffer memory 200...Rotating memory 300...Pad sector table 400...Replacement table

Claims (1)

[Scope of Claims] 1. In a rotating memory control method for reading (and writing) data to a storage unit of a rotating memory, the memory in the storage unit that cannot be stored normally corresponds to the storage unit to be accessed. Prepare a first table that displays the extent of classification, and when reading or writing data to the storage unit, refer to the first table and check if there is a storage section in the storage unit that cannot be stored normally. 2. A control method for a rotating memory, characterized in that data is read or written to the next storage section in the storage unit that can normally store data, skipping that storage section. 2. Scope of Claim 1 In the method, the first table is formed from the second cheesels for the storage unit when selecting the storage unit to access.
-j-j-pull The rotation type is formed from the rotation type memory at the time of initial setting of the rotation type memory, and displays a storage section that cannot be stored normally due to the storage degree of the rotation type memory. 1lill m1 method of type memory. 3. In the method according to claim 1, the first
The table includes a status display section that displays the access status of the storage section included in the storage unit, and when accessing a storage section, refer to the status display section corresponding to the storage section and access the storage section. 1. A first method for suppressing a rotary memory, characterized in that when the access is made, the fact that the access has been made is displayed on the status display section.