Performance neutral isolation of runtime discrepancies in binary code

摘要

Without using a high-level programming language source code, a set of sync points is identified in an initial binary code. The initial binary code is executed at a first system. A value of the user data is captured from a user space of a memory as a baseline of the user data. A set of comparative sync points is identified in a second binary code. During an execution of the second binary code, a second value of the user data from a second user space of a second memory is found to fail in matching the baseline of the user data. An instruction before the comparative sync point in the second binary code is identified as a location of a faulty operation due to the failing.

主权项

1. A method comprising:
identifying, using a processor, without using a high-level programming language source code, in an initial binary code, a set of sync points, a sync point being a place in the initial binary code where a user data will have changed in a memory when the initial binary code is executed; executing, using the processor, the initial binary code at a first data processing system; capturing, using the processor, a value of the user data from a user space of the memory as a baseline of the user data; identifying, using the processor, in a second binary code, a set of comparative sync points, a comparative sync point being a place in the second binary code where the user data will have changed in a second memory when the second binary code is executed in a second data processing system; finding, using the processor, during an execution of the second binary code in the second data processing system, that a second value of the user data from a second user space of the second memory fails to match the baseline of the user data; and identifying, using the processor, responsive to the finding, an instruction before the comparative sync point in the second binary code as a location of a faulty operation.