Memory access instruction instrumentation optimization method in multi-thread application scene

Abstract

The invention discloses a memory access instruction instrumentation optimization method in a multi-thread application scene, redundant instrumentation of a memory access instruction involving a non-shared memory is removed through a static analysis process based on a program intermediate code, and the method specifically comprises the following steps: 1, obtaining the intermediate code; 2, traversing all modules, functions, basic blocks and instructions; 3, acquiring instruction related information, and updating and recording distinguishing information of redundant / non-redundant instrumentation according to corresponding rules; and 4, selectively instrumenting the memory access instruction according to the recorded distinguishing information. According to the method, the problem of instrumentation redundancy in a traditional memory access instruction instrumentation method can be solved, and the purposes of reducing the memory access instruction instrumentation scale and the operationoverhead introduced by instrumentation are achieved.

Description

technical field [0001] The invention belongs to the technical field of performance optimization in memory access behavior monitoring, and specifically relates to a memory access instruction stub optimization method in a multi-thread application scenario. Background technique [0002] In various application scenarios for multi-threaded programs, the memory access behavior of monitoring programs is very important. For example, the detection of concurrency vulnerabilities depends on the tracking and analysis of program access to shared memory. Instrumentation is an effective means of grabbing memory access information, by inserting "probes" (usually code segments containing function calls) into the source program, and throwing out the characteristics of the program runtime through the execution of the probes information (here specifically refers to the memory access behavior information), so as to realize the recording and analysis of the memory access behavior information in ...

AI Technical Summary

Problems solved by technology

However, sampling methods usually need to make a trade-off between accuracy and performance overhead, which may reduce the accuracy while reducing the cost of instrumentation

Moreover, sampling strategies usually depend on specific tasks, and some sampling methods also require additional hardware platform support, so they lack versatility to a certain extent.