JPH0797334B2 - Central processing unit - Google Patents

Description

The present invention relates to a central processing unit, and more particularly to an address trap function for simplifying program debugging in a data processing unit adopting an object-oriented / qualified address system architecture. With respect to the central processing unit.

[Prior art] Computer application systems, which are becoming more and more complicated year by year, pose the problem of increasing the complexity and scale of software built on the computer. Accurate software development is achieved by adopting both the logical approach of developing a system based on it and the practical approach of providing a function for simplifying the debugging of developed software on the computer side. There is an urgent need.

One of the requirements for computer design from both design methodologies is to provide a computer architecture that has little semantic separation from the software system.
In this respect, the qualification address method (capability based add
object oriented architecture (objec)
The computer architecture called "t oriented architecture" is highly evaluated from the old days because the data isolation method based on the object oriented method and the access right check by the qualification address method have a strong affinity with the software design methodology based on the module assembly. Received from inside and outside.

According to this computer architecture, the format of data that is logically cohesive is defined, and then it is first defined as an object type (object type).
type)). A management module is provided for each target type that provides a program package that summarizes the operations and access methods defined for that type. Next, the system is constructed by generating the required number of objects belonging to that type. From the side of the program, the system management program receives the list of objects to be accessed and secures access from the system management program as capability. At the time of execution, the program can access the target by presenting its qualification.

According to this method, the system is beautifully divided / integrated from the three viewpoints of target type, target, and qualification, and the check is performed according to it at the time of execution, so it is easy to make small differences such as changes in the implementation method inside the target. Absorbed by the management module. That is, since the entire system is designed as a set of such packages, the software structure is directly reflected in the environment for program execution, and there is little semantic gap. Therefore, the system construction becomes easy and it is possible to make a system with few errors.

Normally, for one target, the base address in memory in which the target exists, the size indicating the physical size of the target,
It has special information called an object descriptor, which consists of fields that indicate attributes such as the specification of permitted access methods to the data existing inside the target. Since the target descriptor has a link to the management module (or it can be considered as an ID indicating the target type), it is easy to determine what kind of operation is allowed for the target in the first place. On the other hand, a qualification has special information indicating the connection to the target descriptor of the target pointed by the qualification and the type of access permitted through the qualification, and is called a qualification descriptor (capability).

In the qualification addressing architecture, in order for a program to present and access a qualification, a special register that holds the qualification that can be seen by the program is explicitly loaded and the register is specified to access the special register. There have been two proposals, one is to provide the address that you want to access and the other is to automatically search for the qualification without providing it. These two methods are the same in that the access is performed while referring to two sets of information, that is, qualification information and target information to be accessed, although the register configuration that can be seen from the program changes. Therefore, a register having a similar function exists in the computer in terms of hardware (it can be explicitly accessed by the program). Hereafter, explicitly (explicitly) to simplify the explanation
Based on the way the registers are visible to the program,
The method of the present invention can be easily applied to another method, that is, a method in which registers are implicitly designated. In such a qualification register method, a program can explicitly set qualification information in the register to access the target through the qualification. When trying to access the data in the subject through the qualification register,
The access is started after checking whether the access is within the target size, whether the access is legal, etc. Therefore, the qualification register has fields for holding various qualification information and target information. By the way, even though it has the name of qualification register, the information inside this register is not only the information contained in the qualification but also the target information and target type information obtained as a result of accessing through it is set organically. There is.

In the latter debugging practice, each computer provides a function to help various debugging.For example, a single step operation that generates an internal interrupt for each instruction execution and makes it possible to trace the instruction execution, A special instruction is embedded in the instruction segment, and when a flow of instruction execution arrives, an internal interrupt is generated to check the program execution environment after instruction execution. An address trap that causes an internal interrupt when an instruction that attempts to access data existing in a portion other than that portion is executed when the instruction is executed.

Of these, the most efficient function that can support a wide range of debug methods can be considered to be the address trap method. In this method, it is important to set the address range and manage and retain the information, but they are also complicated. As can be easily understood, in the address trap method, the more the number of places where traps can be applied, the better, and the finer the address range is specified.

Conventionally, a computer that adopts the above-mentioned qualified address method has not provided a debugging means such as an address trap. It is considered that this is mainly because in such a computer architecture, each target is so small and simple that even if all accesses to the target are trapped, it does not cause a great load. However, as the system becomes larger and more complex, the system design based on "small object" is being changed to the system design including "large object" (for example, database design). In such a situation, a debugging means such as an address trap is considered to be very effective because it can debug by defining the accessible part and the inaccessible part in one object.

By the way, the function that can divide one target itself and regard it as different targets is also provided as a refinement, but there is a computer provided with this function. I won't mention it here because it's not for debugging.

[Problems to be solved by the invention]

When the address trap is to be applied in the system as described above, the address trap is basically applied to one object. Applying an address trap across multiple objects is not a useful method considering that each object has a different meaning. In such a case, the same function can be achieved by eliminating some of the qualifications given to the program by the system. An easy way to apply an address trap to each object is to provide the address information for the address trap to be applied to each object. At this time, the target descriptor existing for each target is accompanied by some sets of address trap information. This can be thought of as an attribute of each target added specially for debugging. However, in addition to the case where the address trap is applied only to one target, there are cases where it is desired to commonly apply the address trap to a set of targets belonging to the same target type. This is one target
This is an effective means when considering the debugging of a management program in a database system that expresses records. However, since the target itself has an internal structure defined by the target type, it is not preferable that each target has address information separately because there may be duplication. At this time, it is preferable to have address trap information for each target. That is, the address / trap information defined for each target and the address / trap information defined for each target type are separated to prevent duplication. The former is called the target address trap, and the latter is called the target type address trap.

In the target / target type address trap, it is necessary to be able to select the subject that applies the address trap. That is, it is the difference between the debugged program and the debugger. This is a means for allowing the accessing subject to select whether or not to cause an address trap. This is done by having the selection information of whether or not to address trap in the qualification. realizable.

[Means for solving problems]

The central processing unit of the present invention provides target description information including the size and base address of the data as information for managing the data accessed by the program to be debugged under the name of target, and when the program accesses the data. At least for the address trap, which is a technique for generating a trap if the qualification description information for determining whether to permit or disallow the permission at the time of execution with the name of qualification and the address when the program accesses the data A qualification register that stores one address information and is accessed by the program; and obtains the target description information, the qualification description information, and the address information from the qualification register, and accesses the data to the qualification description. Species allowed in information, the target description information, and the address information Access that permits the program to access the data and is the type of access permitted by the qualification description information and the target description information, and is determined to be trapped by the address information. And an access control circuit for generating an address trap, and disallowing access if the access is not of a type permitted by at least one of the qualification description information and the target description information.

〔Example〕

Next, embodiments of the present invention will be described with reference to the drawings.

1 and 2 are a data access diagram and a functional block diagram, respectively, showing an embodiment of the present invention.

FIG. 1 shows a data access method in the architecture which adopts the object-oriented computer system technology adopting the qualified address system. The one indicated by 110 is the target. Reference numeral 150 is a target descriptor having information for managing the target. As can be seen from the figure, the address / trap information is attached. The one indicated by 200 represents the object type to which the object belongs. As you can see from the figure,
Address trap information is attached. What is indicated by 300 is a management module which is a program group for managing the target type. What is shown at 400 is qualification. It is in the form of a credential descriptor. You can see flags for address trap selection control. By setting this qualification in the qualification register indicated by 10, the program can access the data in the subject pointed to by the qualification through the management module. When the program loads the qualification into this register, the information from the target descriptor, target type, and qualification descriptor is combined and stored in the qualification register 10 according to the method shown at 600.

FIG. 2 is a block diagram for checking access rights and detecting address traps when accessing data through the qualification register 10 during program execution. The designated 10 is the qualification register. Reference numeral 100 is the base address of the object accessed through the qualification, 101 is the field indicating the size of the object, 105 is the control field of other information, 103 Information indicated by 104 and 104 is information for the address trap. Although only two address trap information fields are depicted in this example, it will be clear that increasing the number is easy. A control field for controlling the address trap is also included in this field.

In the conventional qualification register system, such a field related to the address trap does not exist, which is a difference from the present invention.

What is specified by 20 is the access right check circuit. The access address register indicated by 30 is 201
The size comparison circuit of the upper limit check pointed to by (1) checks whether the access is valid within the target size defined in the field 101. The combinational circuit defined by the access right checking circuit 20 checks whether or not the access is violating the access right defined in the field 102. Comparing circuits for referencing the address trap information defined by 103 and 104 and determining whether to cause an address trap for the access are respectively address trap comparing circuits.
203 and 204. 205 is an address trap control circuit that defines whether to cause the address trap.

That is, by storing the address trap information in the qualification register 10 as well, the address trap for each target / target type can be realized by referring to only the information in one register. Such a qualification register 10
Can be provided in the central processing unit as many as required by the architecture. However, the address / trap control circuit 205 may be one circuit set described in FIG. Also, as mentioned earlier, this qualification register 10 implies not only architectures that are explicitly (explicitly) accessed by the program, but registers with similar functionality that are not visible to the program when data is referenced through the qualification. It can also be applied to the architecture of the method of loading dynamically (so-called caching).

〔The invention's effect〕

As described above, the present invention is superior to the conventional system in the computer system having the access means for accessing the data designated by the qualification register, by having the function of the address trap by the configuration of the qualification register and the access control circuit. This has the effect of facilitating program debugging.

[Brief description of drawings]

1 and 2 are a data access diagram and a functional block diagram, respectively, showing an embodiment of the present invention. 10 …… Qualification register, 20 …… Access right check circuit,
30 …… Access address register, 100 …… Base address field, 101 …… Size field, 1
02 ... Access designation field, 103,104 ... Address / trap information field (ATI), 201 ... Size comparison circuit, 203,204 ... Address / trap comparison circuit, 205 ...
… Address trap control circuit, 110 …… Target, 150 ……
Target descriptor, 200 ... Target type, 300 ... Management module,
400 …… Qualification, 600 …… Composite information of target descriptor, target type, and qualification.

Claims (1)

[Claims] 1. Target description information including the size and base address of the data as information for managing the data accessed by the program to be debugged under the name of target, and permission / inhibition when the program accesses the data. At least one address for an address trap, which is a technique for generating a trap if the qualification description information for determining permission with the name of qualification at the time of execution and the address when the program accesses the data A qualification register that stores information and is accessed by the program; and obtains the target description information, the qualification description information, and the address information from the qualification register, and accesses the data to access the qualification description information, With the type of access permitted by the target description information and the address information In some cases, the program is allowed to access the data, the type of access permitted by the qualification description information and the target description information, and the access determined to be trapped by the address information An access control circuit for generating an address trap and disallowing access if the access is not of a type permitted by at least one of the qualification description information and the target description information. Processing equipment.