JPH01223673A - Data recording control system - Google Patents

Abstract

PURPOSE:To surely execute the erasion processing of a defective record area by executing the erasion processing of the defective record area by only the decision of a controller side, when a track overrun is generated and a write failure record area is generated. CONSTITUTION:When a track overrun is detected, a head SN of a C part 212 of Rn 211 which has become a track overrun and the defective record is stored as SNa. Subsequently, SN of a C part 208 is calculated, and denoted as SNb. SNa and SNb are compared, and when they coincide, it is recognized as the defective record caused by a track overrun, the erasing operation of the C part 212 of the defective record is executed. Next, the erasion processing of a K part and a D part is executed. After the erasion processing to 1X201 is ended, an error end report is executed. In such a way, by reading continuously each record, while waiting for the rotation of a magnetic disk, and confirming the physical position on a track, the position of a head on the track can always be confirmed in the course of a repositioning operation.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a data recording control method for a magnetic disk subsystem, and in particular to a method for reducing the load on an O8 error recovery program for track overrun failures and for data storage. The present invention relates to a data recording control method suitable for improving reliability.

[Conventional technology]

FIG. 6 shows a general system configuration of a data recording system including a magnetic disk storage device (rotary storage device).

Central Processing Unit
g Unit (hereinafter abbreviated as CPU) 101 and channel (hereinafter abbreviated as CHL) 102, a magnetic disk controller (Disk Controller)
, hereinafter abbreviated as DKC) 103 is connected, and a magnetic disk storage device (Disk Storage Con) 103 is connected.
troller.

(hereinafter abbreviated as DSC) 104 is connected. DSC-10
4 includes n units (0 to n-1) of magnetic disk storage devices (
Device (hereinafter abbreviated as DEV) 105 is connected.

Figure 2 shows CK, which is a general variable length recording method, on the magnetic disk medium (rotary storage medium) in the DEV 105.
It shows the track status of one rotation, that is, one track, according to the D format. Furthermore, a track in which a defective record is created due to a track overrun is shown. FIG. 3 shows an example of a command chain that caused a track overrun.

In Figure 2, there are index markers (hereinafter abbreviated as IX) that indicate the beginning position of each track (in the figure, there are two IX
201 (indicates the same thing), there is a home address (hereinafter abbreviated as HA) 203 with a gap 202 in between, and record 0 (hereinafter referred to as R) with a gap 202 in between.
(abbreviated as 0) 204° record 1 (hereinafter abbreviated as R1) 2
05. There are arbitrary n records such as record 2 (hereinafter abbreviated as R2) 206 up to records R,,-, (hereinafter abbreviated as R7-1) 207. Each record is
Count section (hereinafter abbreviated as C section) 20.8. Ki part (hereinafter abbreviated as part) 209. Data section (hereinafter abbreviated as the following section) 21
It is composed of 0 fields. Furthermore, IX
Markers are recorded in advance on each track, but in the variable length recording method, not only the K and D sections but also the 0 section are updated by erasure processing.

Also, Ro 204. Part 209 like Rz 206
Records that do not contain are also allowed.

FIG. 3 shows an example of a command chain for creating a new record on a track. Seek command (SE
(EK, hereinafter abbreviated as SK) 301 is the cylinder address,
Information indicating the head address is received from the channel, and the head is positioned on the track of the specified cylinder address and head address. SEARCHID EQUAL (SCHI)
D EQ) 302 receives information indicating a part of ID (ID part = cylinder address, head address, record number, hereinafter abbreviated as ID) from CHL 102, and selects 0 where the same ID is written. Section 208 is searched. The tick command (TiC) 303 is
If D does not match, try SCHID EQ30 again.
Issue 2.

Write Sea Cape Command (WRITECKD)
, hereinafter abbreviated as WRCKD) 304 uses the data transferred from the CHL 102 as write data and writes new records in the order of 0 part, 2 part, and D part.

Next, an example of a case where a track overrun occurs is shown in the second example.
This will be explained using FIG. 5K30LSCHID
EQ302. Using the TiC303, the head was positioned at record R, 207 (one place before the target record) on the 8-track, and when a new record, record n (hereinafter abbreviated as Rn) 211, was written using the WRCKD304, the writing was completed. Since the number of data bytes was greater than the number of remaining bytes on the track, lX201 was detected during execution of the write operation, and the write operation was abnormally terminated.

R, 211 remains on the track as a defective record that has been written only halfway.

Such failures are commonly referred to as track overruns (
Track 0verrun) or invalid track format (Invalid Track F)
ormat).

If a track is accessed while a defective record due to a track overrun remains on the track, an error will occur when reading the defective record.

An error in reading the record becomes an error not only when a read command is performed on the track, but also when a search command is performed on the track.

In the conventional technology, when a track overrun is detected, an error completion report is sent to the CHL 102, and a sense byte (24 bytes of data indicating error details) is sent to the CHL 102.
In response to an overrun error report, the operating system (Operating S
system (hereinafter abbreviated as O8) erases defective records using an error recovery program. Figure 4 shows an example of a command chain for erasing bad records issued by an error recovery program.
The process will be explained using FIG.

Attach the head to the truck with 5K401 (stand upright,
Read home address (READ HA, hereinafter referred to as R)
(abbreviated as D HA) 402 and HA20 at the beginning of the track.
Position it immediately after 3. Furthermore, SCHID EQ403
and R, ! on the track by TiC404. 2-07
Positioned in The erase command (ERASE, hereinafter abbreviated as ER3) 405 is a command to receive arbitrary data from the CHL 102 and then erase it until the track head mark lX201 appears.
Since the command is received while positioned at -,207, lX201 starts immediately after the 0 part 210 of R,+207.
The erase operation is performed until Through these processes, the defective record R, 211 remaining on the track is erased.

In addition, as related to the above-mentioned prior art, Japanese Patent Application Laid-Open No. 58
-48161 publication is mentioned.

[Problem to be solved by the invention]

In the above-mentioned conventional technology, when the magnetic disk control device learns of the occurrence of a bad record (track overrun), it first reports this to the host device, and the host device that receives this report sends an O8 to check the bad record. An erase command is issued, and in accordance with the command, the magnetic disk system 41U erases the defective record. Therefore, according to the prior art, error recovery processing for track overruns, that is, deletion processing of bad records, is performed based on an input/output instruction by the error recovery program of O8, which causes the following problems. be.

1. Disc system? 11 After the track overrun error is reported from the device to the host device, - If the host device does not issue an I/O command to delete the bad record for some reason, or this I/O command is still issued. In the previous case, the bad record remained on the track. If, for example, another CPU issues a new data read command or the like on a track where a defective record remains, a read error will occur in the defective record. This is because once an error report has been made, the storage device is in a state where it can accept activation from other CPUs.

2. In order to implement error recovery processing by O3, a new input/output instruction is required, which becomes a burden on O8.

Generally, processing such as erasing data is important, and conventionally it has been thought that instructions must be sought from the host device each time. However, the inventors of the present invention have focused on the fact that erasure processing for defective records such as track overruns can be determined only by the control device.

Therefore, one object of the present invention is to provide a data recording control system in which a defective record remains on a track when the O8 does not perform defective record erasing processing or until a defective record erasing command is issued. The object of the present invention is to prevent a read error from occurring due to a new access to the track in the current state.

Another object of the present invention is to reduce the burden of error recovery processing by the O8 in a data recording control system.

[Means to solve the problem]

In order to achieve the above object, the data recording control method of the present invention basically detects the occurrence of a track overrun when writing data in a rotating storage device based on a write command from a host device. Instead of immediately reporting what has happened (from the control device to the higher-level device) as in the past, and then erasing it using the input/output command from the higher-level device using the O8 error recovery program, the control device first The defective record portion due to the overrun is erased and removed based on the judgment, and the track overrun is reported to the host device after the erase operation is started or after the erase is completed.

Furthermore, the defective record portion is erased under control of the control device by utilizing repositioning of the magnetic head while waiting for rotation of the rotary recording medium.

In a specific example, as a means for repositioning the magnetic head, the physical position (segment number) of the defective record that caused the track overrun is stored on the control device side, and the stored physical The target position is compared with the physical position (segment number) of successive records read during one rotation of the recording medium, and a match is detected.

[Effect]

The effect based on the above configuration will be explained. 'After the occurrence of a defective record resulting in a track overrun, the erasing process for the defective record portion is completed while the recording medium is waiting for one revolution.

This operation will be described in detail below.

Immediately after a track overrun is detected, CHL
102 is maintained.

At the same time, the physical position of the bad record on the track is recorded as a target value for repositioning to the bad record. This physical position is based on a predetermined point on the track (usually IX 201) and is indicated by the absolute distance to each record.

Next, each record is successively read from the HA 202 at the beginning of the track, the physical position of each record is recognized,
The physical location of the next record is derived, for example, by calculation. In variable length recording, the HA and 0 sections have a fixed length, but the K and D sections have variable lengths, and their lengths are usually managed by the C section, so the length of each field (bytes) By calculating the length) and the interfield gap length (byte length), the physical position of the next record can be calculated. By comparing the derived physical position of the next record with the physical position of the bad record that caused the track overrun, searches for the beginning of the record at the same physical position as the bad record. do. When detected, the record is recognized as a defective record created by a track overrun, that is, R, 211, and from then on, erasure processing is performed up to lX201. After the erasure processing is completed, a track overrun error report is sent to the CHL 102. .

As described above, the control device does not report an error immediately after detecting a track overrun, but maintains the connection state with the CHL 102 until an error report is made after the deletion process of the defective record is completed. does not accept new invocations of . Therefore, operations such as read and search are not performed while a defective record remains on the track (and a read error does not occur).

Since no data is transferred between the DKC 103 and CHL 102, from track overrun detection to track overrun error reporting after the defective record deletion process is completed, this information is necessary for conventional O3 error recovery processing. , the number of remaining bytes indicating how many bytes of unwritten data remain when a track overrun occurs is guaranteed.

Therefore, there is no need to make any changes to the interface with the conventional O8, and operation is guaranteed even under the conventional O8.

Furthermore, erasure of defective records can be reliably performed only by control based on the judgment of the control device (based on the micro program), without depending on the O3 recovery program of the host device.

〔Example〕

Embodiments of the present invention will be described below with reference to the drawings.

First, let's look at an example of how to represent the physical position on the track.
Illustrated with diagrams. The track is evenly divided into units of several bytes with lX201 as the base point. This one unit will be called a segment. By counting the number of segments from IX 201, the physical location on the track is determined. This value is referred to as the segment number (hereinafter SN
(abbreviated as ). Each field on a track recorded in the CKD format, that is, the HA, C section, NI section, and D section, is defined so that it always ends at a segment boundary (as in section D, the end of data does not necessarily end at a segment boundary. Sometimes they don't come, but at that time,
Add an auxiliary pattern to match the end of the data to the boundary. ), the position of each field is represented by SN. In FIG. 5, the number of segments from lX 201 to the rear end of 0 part 208 is SN. If it is determined that the SN at the rear end of each field of each record is written in the HA 203 and the 0 part 208 of each record, the physical position of each record on the track is indicated.

Next, the configuration and operation of the data recording control method according to the embodiment of the present invention will be explained with reference to the flowchart of FIG. 1, the track format of FIG. 2, and the system configuration diagram of FIG. 6.

In FIG. 1, when a track overrun is detected in step 601 (the detection of this track overrun is the detection of tX immediately after the defective record R7, that is, the index representing the beginning of the track, that is, the track overrun is detected, and the defective record is detected). 0 part 21 of Rfi211 which became
Store the first SN of 2. In this example, for convenience, this is SNa
shall be. In step 603, a read operation of the HA 203 is performed to position the head at the beginning of the track.

Next, in step 604, the SN of the Vt<c section 208 is calculated. For convenience, this is referred to as SNb. In step 605, SNa and SNb are compared. As a result of this operation, if the two are not equal, the record is a bad record R,l
It is recognized as a good record before 211, and step 60 is performed.
6 reads part 0 208, and reads part 20 in step 607.
In RECORD where the K part does not exist, the D part is immediately skipped. ), the record is positioned immediately before the next record, and the step returns to 604. By repeating this process and searching for the SN at the beginning of successive records, defective records are detected.

When SNa and SNb match in step 605 (
As mentioned above, the SN at the end of each field of each record is written in the C part of each record, so when you reach the position just before the C part of the bad record, you can find the SNa of the next bad record. Matching of SNb can be confirmed. ), the record is recognized as a defective record due to track overrun, and in step 609, the 0 section 212 of the defective record is erased. Continue with step 6
10, erase the K part and the D part (if the K part does not exist, immediately erase the D part), and lX201
After the erasing process is completed, in step 611 the CHL
An error completion report is sent to 102.

According to this embodiment, when repositioning a defective record caused by a track overrun, each record is successively read while waiting for rotation of the magnetic disk, and the physical position on the track is determined. By checking,
Repositioning has the effect that the position of the head on the track can always be accurately recognized during operation. Furthermore, since the search for the corresponding defective code is performed by comparing the physical positions on the track, accurate positioning at the beginning of the target record is possible. Therefore, good records other than bad records will not be erased by mistake (the authenticity of existing data is guaranteed, and bad records will be erased during one rotation of the magnetic disk from the time the bad record is written). , everything is erased accurately.

In the above embodiment, an error report is sent to the CHL 102 after all of the deletion processing of the C, 2, and D parts of the defective record is performed.
After performing the erasing process of CHL in step 611,
102, and the erasing process of the K section and the D section is a format writing operation (format writing, which is a writing method that updates not only the N section and the D section but also the entire C section. After the writing up to record R8 is completed, the drive device then uses a signal from the control device to erase all subsequent parts up to IX. Conceivable.

In this way, the error completion report can be made as soon as the 0th copy of the defective record has been erased, without having to wait until the deletion process is completed up to lX201, so if the D section of the defective record is extremely long, This has the effect of allowing faster error reporting. Even in this case, the erasing process is not performed by the error recovery program by the O8 from the host device (but is performed by the 'ζ driving device itself using a signal from the control device, so the erasing is reliably performed).

[Effects of the Invention] As described above in detail, according to the data recording control method of the present invention, when a track overrun occurs and a writing defective record area is generated, the erasing process of this defective record area is performed. , O8 of a conventional host device
Since the deletion process is performed solely based on the judgment of the control device without any intervention, there is an effect that the deletion process of the defective record area is performed reliably. In addition, since the track overrun is reported to the host device after the defective record area has been erased, the O8 of the host device does not need to issue a command to delete the bad record, reducing the burden on the O8. Ru. Furthermore, the connection status of the storage device to the higher-level device is maintained until a track overrun is reported to the higher-level device, so other CPUs can
It is no longer possible to accept booting to the storage device from other sources, resulting in excellent effects such as improved reliability.

[Brief Description of the Drawings] Fig. 1 is a flowchart for explaining the configuration and operation of an embodiment of the present invention, and Fig. 2 shows tracks according to the CKD format in a general variable length recording method on a magnetic disk recording medium. Figure 3 is a diagram showing the new record creation command chain, Figure 4 is a diagram showing the bad record deletion command chain, Figure 5 is an explanatory diagram of segment numbers, and Figure 6 is a diagram showing the magnetic disk storage device. 1 is a configuration diagram of a general data recording system including: FIG. 101-・-1---Central processing unit, 102-・～・・
--- Channel, 103 --- Magnetic disk control device, 104 --- Magnetic disk connection device, 105-
-Magnetic disk storage device, 201-- Index marker, 20 S''-- Gap, 203-- Home address, 204-2
07------Record, 208---Counter section, 209---Key part, 21 (1-------
-Data section, 211---Bad record created by track overrun, 212--
・Count part of defective l-1', 301 to 304.40
1 to 405 --- Command, 601 to 611 ---
---Step. -Figure 1 Area 1 Figure 3 Figure 4 Figure 5

Claims (1)

[Scope of Claims] 1. A central processing unit, a channel device, a control device, and a rotary storage device controlled by the control device and connected to the channel device, wherein data from the central processing device is controlled by the control device and connected to the channel device. In a data recording control method in which data is recorded in a variable length manner in a rotational direction of a storage medium of the rotary storage device according to a write command, on the control device side, an amount of write data specified by the data write command; detecting means for detecting the occurrence of a write defective area when the write error area exceeds the allowable storage capacity in the rotational direction of the storage medium; and erasing means for immediately repositioning the rotary storage device in the write-defective area and erasing the write-defective area when the write-defective area is detected. 2. The device further comprises means for reporting the occurrence of the write failure area from the control device to the channel device when the erasure of the write failure area is completed by the erasing unit. 1. The data recording control method described in 1.