| 摘要 |
In accordance with the present invention, computer implemented methods and systems are provided that allow an application to automatically recover from software failures and attacks. Using one or more sensors, failures may be detected in the application. In response to detecting the failure, the porti on of the application's code that caused the failure is isolated. Using the inp ut vectors that caused the failure, information regarding the failure (e.g., th e type of failure), a core dump file (e.g., stack trace), etc., an emulator- based vaccine that repairs the failure is constructed. In response to verifying that the vaccine repaired the failure, the application is automatically updated with the emulator-based vaccine without user intervention. Application community features that efficiently use the resources available in software monoculture is also provided. An application community may be defined that includes a plurality of devices and the application's code maybe divided into smaller portions of code, which are assigned to each of the plurality of devices for monitoring. Each device als o notifies the other devices of the failure.
|
