JPS63148349A - Cache memory positioning system - Google Patents

Abstract

PURPOSE:To perform the positioning of an address corresponding to a designated sector in a short time, by making a pair of the sector value of a track on a magnetic disk and the address on a cache memory corresponding to the sector value, and storing it at every access. CONSTITUTION:When a set sector command is supplied, its sector value St is stored in the register 41 of a data memory 4. Next, the sector value (m) and a memory address P at a preceding access time are read out from the prefix part of the track corresponding to the St of the cache memory 1, and are set at registers 42 and 43 respectively. And the St is compared with the (m) at an arithmetic circuit 7 and a decision circuit 8, and by setting the P when it is St>=m, and the leading record address of each track when it is St<m, as the start point of address calculation corresponding to the St, and finding the sector value from each record in order, the positioning of the address on the cache memory 1 corresponding to the St is performed.

Description

[Detailed Description of the Invention] [Summary] When a specified sector value of a magnetic disk is given to a disk cache of a magnetic disk, and when finding the corresponding address position on the disk cache, Using information in pairs of sector values and their corresponding addresses.

By determining the address of the record that is the starting point of the current address calculation according to the magnitude relationship between the current specified sector value and the previous sector value, the time for locating the address is shortened.

[Industrial application field]

The present invention relates to a disk cache of a magnetic disk used in a computer, and particularly relates to a cache memory positioning method corresponding to sector values [prior art]. The storage state on the disk cache memory corresponding to the track is shown in FIGS. 4(al) and 4(b).

In FIG. 4(a), the tracks of the magnetic disk.

A certain area (word length) starting from the index mark position
Each sector is given a sector value (number), and on the track, following the home address section HA, there are records RO, which consist of a count section C and a data section.
R1, R2... are recorded.

As shown in FIG. 4 (al), since the data length is variable according to the variable length recording format, the storage state of the disk cache memory corresponding to the track of this magnetic disk is As shown in Figure 4 (b), the recording gap G (Figure 4 (al) is deleted and each record is stored sequentially at consecutive address positions in the memory without any gaps), which is always provided in magnetic disk recording. There is.

There is a cent sector (Set 5ector) command as an access from the CPU to the record of the track recorded in this disk cache memory.

This command is given by specifying the sector value on a specific track on the magnetic disk.

In a conventional disk cache control device, when executing this command, since the start address (HA storage location) of the track in the disk cache memory is obtained from a table, etc., the start address and the start sector value 0 are used. Find the memory address corresponding to the specified sector value.

For this purpose, conventionally, when reading the memory at an address position after a certain word length (HAO word length) from the head position address of the memory, the result is as shown in FIG. 4(b).

There is a record RO count section, and since part of its contents includes the data length of the record, the memory address of record R1 can be obtained as a primary source.

If the sector value specified by the cent sector command is m, each recording section H from the beginning position of the memory address
A, RO-C, RO-D, R1-C, R1-D,...
, and the total word length of the gap G between each recording section are added, and this is divided by the word length specific to one sector to convert it into a sector value.

If the result is smaller than m, read the cache memory using the memory address of the next record (obtained by adding the data length of the count section of the current record to the memory address of the previous record), and add the data length and gap length. is added, converted to a sector value, and compared with 2m. When the value is equal to or greater than m, the address at which the memory address at that time corresponds to the specified sector value is determined.

This completes the positioning.

[Problem that the invention seeks to solve]

As mentioned above, according to the conventional method of positioning addresses with respect to specified sector values, it is necessary to reach the specified sector value by tracing each record in order from the beginning of the track in the cache memory, so it is difficult to position the address to the record 5 records later. The problem was that it took time.

[Means for Solving the Problems and Their Effects] In order to solve the above problems, the present invention uses the previous cache memory address obtained by the set sector command and the specified sector value in the cache memory track. When the sector value stored in a specific position is given and a secondary set sector command is given, the sector value stored in the specific position is compared with the specified sector value in the command, and if the specified sector value is greater than the stored sector value, In some cases, by tracing the records starting from the address stored at a specific location to find the address corresponding to the specified sector, this method saves time compared to the conventional method of always tracing records from the beginning of the track. It is.

The basic configuration of the present invention is shown in FIG. Also, Figure 2(a)
, (b) respectively show the recording arrangement of tracks on the magnetic disk and the corresponding storage arrangement on the cache memory according to the present invention.

In FIG. 1, the area within the solid line frame is a disk cache control device, 1 is a cache memory, 2 is a data register, 3 is an address register, and 4 is a plurality of Q registers 41 to 46.
Data memory consisting of etc.

5 is an X register, 6 is a Y register, and 7 is an arithmetic circuit.

8 is a discrimination circuit, 9 is a control circuit 910 is a program memory, 11.12 is a bus, control operations are performed by a program, and a control circuit 9 controls the arithmetic circuit 7 and gates between the circuits.

The arrangement of the track data of FIG. 2 (, 9) stored in the cache memory 1 is shown in FIG. 2 (bl).

FIG. 2 (of the storage arrangement of the cache memory in BL).

The brifix portion is unique to this invention.

The bfix section is used as an area for storing a set of a specified sector value given by a command at the time of access and a cache memory address determined by the sector value when the track in question is accessed. .

The contents of this brifix section are rewritten every time there is an access, and the contents are used to make the first access more efficient.The brifix section is provided at a specific data position of each track. , in this example, is located before the home address portion of the track.

〔Example〕

Next, an embodiment of the present invention will be described according to the operational flow shown in FIG. In addition, Fig. 1 and Fig. 2 +8), (b
) is referenced.

In step (2), a set sector command is given from the cPυ side (not shown).

The designated sector value is stored in the register 41 of the data memory 4. In this case, along with the designated sector, the track designation and the start address of the track in the cache memory are also given, but the description will be made assuming that the address of the start position of the track in the cache memory is O.

By receiving a set sector command.

The control operation that has been started is carried out in the second step (2).

The bfix part (in Figure 2) of the relevant track in cache memory 1 is stored in address register 3 (the address one word before address 0, where k is the number of words in the prefix).
Read by address. The contents of the bfix part are transferred to 4 of cache memory 1 via data register 2.
2.43 as a sector value m and a memory address P, and the arithmetic circuit 79 determines whether the contents are other than O (whether or not a value is set).

If the value is set, it becomes clear that the data of any record in this track has been accessed before, and in the first step ■, the sector value (assumed to be m) at the time of the previous access is determined. The contents of the register 42 that holds , and the record 41 that holds the designated sector value St.
The content of 1 arithmetic circuit 71 is discriminated by discriminating circuit 8.

If it is mm5t, it means that the desired specified sector value matches the sector value when accessed last time, and the desired memory address is the memory address P of the bfix stored in 43 of cache memory 1 itself. So I'll end it here.

In step (2), if it is determined that m<St, it is understood that the specified sector value St this time is later on the track than the sector value (m) at the previous access.
In order to start address calculation based on the sector value m at the time of the previous access and the memory address P at that time, in step (3), the values of m and P are transferred from the cache memory 1 to the X register and the Y register, respectively.

Step (2) is to determine the end condition of the program loop to be located, and it is checked whether data representing an index has been written in the memory location.

In this case, since no index exists at the address P, control is transferred to the first step, where the calculation of the second order equation is executed.

X+ (L i +G) /5n-X −(1)Y+
L i →Y ...... (2) Equation (11 is the previous sector value m in the X register 5.

One sector word length (
This represents the operation of adding the value obtained by dividing the fixed word length (fixed word length) by Sn (the value obtained by converting the record word length and the gap length into the number of sectors) and storing the result in the X register 5.

Also, equation (2) is expressed as address P, which is the content of Y register 6.
By adding the data length of the record to and storing the result in the Y register 6, the address of the record next to the record at the address P is obtained.

Recording gap length G used for calculation of equation (11, (2+)
and 1 sector word length Sn. When the register calculation in the data memory 4 is completed, the first step proceeds to step 2, and the contents of the X register 5 storing the sector value of the next record obtained by calculation and the specified sector value St are The contents of the stored register 41 are compared, and if X<St, it is found that the record corresponding to the specified sector value is located at a later position, so the process returns to step 3, and from there the operation described above is performed. It will be done.

When it is determined in step (2) that X≧St.

Since the sector value of the X register 5 matches or exceeds the designated sector value St, the memory address value stored in the Y register 6 at this time is the desired memory location corresponding to the designated sector value.

The operation ends here.

Now, returning to step ■, if there is no data in the prefix field, that is, if no cent sector command has arrived for that track before then, proceed to step ■, and read the sectors sequentially from the first record of the track. To calculate the value, set the value O representing sector O in the X register 5, and set 0, which is the start address of the track in the cache memory, in the Y register 6. Perform the following actions in sequence.

Also, if it is determined to be mast in step ■,
Since the sector value m at the time of the previous access is located after the specified sector value this time, in step The same process as when there is no data in the prefix part is executed in step ■.

In step (3), when the contents of the Y register 6 are written to the address register 3 and read from the cache memory, if the contents are not record data but an index is detected, the process moves to the next step. In this case, in the process of following records one after another, the index (HA) that is the start position of the track in question is detected, which means that the previous record has the word length up to the final sector position, and the track The next record position will be the first position on the track of the vL disk, as shown in Figure 2 (al), so the memory address to be sought is the first memory address of the track, so set O to the Y register 6. Set and exit.

In this way, the address on the cache memory corresponding to the specified sector given by the set sector command is obtained in the Y register 6, and the positioning is completed.

After the positioning is performed, it is necessary to write the sector value specified by the current command and the obtained memory address into the prefix portion of the cache memory.

〔Effect of the invention〕

According to the cache memory positioning method of the present invention, compared to the conventional method of always sequentially tracing records from the head position of a track, when the setter value is the same as or larger than the previous positioning, especially when positioning to a later record, The number of operations required to locate the specified sector can be reduced.1
Positioning time can be shortened.

[Brief explanation of the drawing]

FIG. 1 is a diagram showing the basic configuration of the present invention, FIG. 2 (al is a diagram showing the recording format on the track of the magnetic disk, and FIG. 2(b) is a diagram showing the storage arrangement of the cache memory according to the present invention. , FIG. 3 is a diagram showing the operation flow of the embodiment of the present invention, FIG. 4(a) is a diagram showing the recording format on the track of a conventional magnetic disk, and FIG.
FIG. 3 is a diagram showing a storage arrangement of a cache memory for track data according to a conventional method corresponding to item a). In Figure 1. l: Cache memory 2: Data register 3 Near address register 48 Data memory 5.6: X, Y register 7: Arithmetic circuit 88 Discrimination circuit 9: Control circuit 10 Niprogram memory

Claims (1)

[Claims] In a disk cache for a magnetic disk in a variable length recording format, a sector value of a track on the magnetic disk and a corresponding address on the cache memory are paired to correspond to each track on the cache memory. The sector value and the specified sector of the set of information stored in the specified position of the corresponding track are stored at a specific position each time each track is accessed, and in response to an access to a specified track of the cache memory that specifies a sector value. If the specified sector value is the same as or larger than the stored value, the stored address as a set is used as the starting point for calculating the address corresponding to the specified sector value, and the specified sector value is is smaller than the stored sector value, the first record address of each track is used as the starting point for calculating the address corresponding to the specified sector value, and the sector value is sequentially calculated from each record. A cache memory positioning method characterized by positioning at the upper address.