Apple Inc.

Alan W Lillich

Erik L Eidt

Sean R Parent

Eric P Traut

Bruce K Jones

Jean-Charles D Mourey

Jeffrey Robert Cobb

Wayne N Meretsky

Apparatus for executing a plurality of program segments having different object code types in a single program or processor environment

Relocation table entries in a executable object code file are interpreted as relocation instructions rather than as individual specifications for a particular respective relocatable information item. An abstract machine is provided for interpreting the relocation instructions and performing various relocation operations and various control functions for the abstract machine, in response to the relocation instructions. Certain variables contain information which is referenced and updated in response to certain types of the relocation instructions, thereby obviating the need to include such information as part of each relocation instruction. Certain of the relocation instruction types can also specify a particular relocation operation to be performed on a run of n consecutive relocatable information items, where n is specified as part of the relocation instruction. Certain of these relocation instruction types also consider each information item as including two or more individual relocations of the same or different types to perform, thus effectively specifying a run of n repetitions of a pattern of relocations. Other types of relocation instructions are also made available.

Relocatable object code format and method for loading same into a computer system

The present invention teaches a variety of methods, apparatus and data structures for providing data driven patching. According to one embodiment, patches are stored in a known format in a discernible location. In the described embodiment, each fragment code may have a corresponding patch library. This enables the patches to be located and analyzed in a quiescent state. In a method aspect of the present invention, the operating system, or a separate utility program, can evaluate and selectively add patches. Therefore, the present invention introduces a patch integrity validation layer into the patching process. In another method aspect, the invention teaches evaluating the patches in a quiescent state whereby the patches introduced by a program or a combination of programs may be exhaustively evaluated prior to execution.

Methods, apparatus, and data structures for data driven computer patches

Relocation table entries in a executable object code file are interpreted as relocation instructions rather than as individual specifications for a particular respective relocatable information item. An abstract machine is provided for interpreting the relocation instructions and performing various relocation operations and various control functions for the abstract machine, in response to the relocation instructions. The abstract machine maintains certain variables containing information which is referenced and updated in response to certain types of the relocation instructions, thereby obviating the need to include such information as part of each relocation instruction. Certain of the relocation instruction types can also specify a particular relocation operation to be performed on a run of n consecutive relocatable information items, where n is specified as part of the relocation instruction. Certain of these relocation instruction types also consider each information item as including two or more individual relocations of the same or different types to perform, thus effectively specifying a run of n repetitions of a pattern of relocations. Other types of relocation instructions are also made available. The file also contains a hashed symbol export table and pattern-initialized data expansion instructions.

Relocatable file format and method and apparatus for creating and

The present invention discloses a variety of methods and apparatus for providing patches within a computer operating system. A patch structure in accordance with one embodiment of the present invention includes a patch block which serves to link the patch structure into a patch chain, and a patch which contains the desired functionality of the patch structure. The patch chain includes a root patch structure and a final patch structure. The root patch structure has a root patch block and a given function which has the root functionality. One aspect of the present invention teaches a method for integrating patches into a computer operating system including the steps of obtaining a new patch to be associated with a patch chain for the given function, determining whether a patch chain is already created in the operating system for the given function and creating a root patch block pointing to the given function as the start of a new patch chain if a patch chain has not already been created for the given function, creating a new patch block for the new patch, and inserting data into the new patch block to cause the new patch block to point to the new patch.

Method and apparatus for patching operating systems

The invention is a method and apparatus for verifying compatibility between components of a system which share a client-provider relationship. Briefly, according to the invention, a current version of a provider and a compatibility range are defined for each of a version of a client and a version of a provider. A version of a provider specifies an oldest implementation provider and an oldest definition provider. When a client is linked with a particular version of a provider it stores an identification for that provider, a current indicator for that version of the provider, called a definition provider, and the oldest implementation provider. At runtime, compatibility checks are performed between a client and available versions of the provider(s), called implementation providers, with which it has been linked. For each available version of each type of provider compatibility exists with the client in three situations. First, if the definition provider and the implementation provider are the same version of that provider, then the client and provider are compatible. Second, if the definition provider, i.e. the version of the provider linked with the client, is newer than the implementation provider, then if the version of the implementation provider is no older than the oldest implementation provider specified in the client, the two are compatible, otherwise they are incompatible. Third, if the definition provider is older than the implementation provider, then if the definition provider is no older than the oldest definition provider specified in the implementation provider, then the two are compatible.

Method and apparatus for determining at execution compatibility among

The invention provides a method and apparatus for switching between execution of a plurality of object code types having different conventions for invoking program procedures and performing stack manipulations. The invention may also be used to switch between different calling conventions within a single object code type. Briefly according to the invention, a computer system comprises a routine descriptor, a stack switch frame, a mode switching mechanism for switching from a first processor, code or calling convention type to a second processor, code or calling convention type and means for executing instructions in various code type codes. A routine descriptor describes a program or code segment and its code type and calling conventions. A routine descriptor contains, among other information, a 'mixed mode' field which is set to a specific, predetermined value such as a value indicating an instruction which is not legal in the runtime environment of a first processor, code or calling convention type. When that instruction is encountered, control is transferred to the mode switching mechanism. A routine descriptor also contains a 'procedure information' field which is set to a value indicating the convention for invoking a program segment and performing appropriate stack manipulations. When a routine calls a routine having a different stack model, the mode switching mechanism uses a stack switch frame to provide a transition between the two different stack types.

Growing community of inventors

Los Gatos, CA, United States of America

Alan W Lillich