Method for reducing processor cycles used for global offset table address computation in a position independent shared library

摘要

A process for compiling shared library source code into libraries in which the global offset table location is calculated only when a cross module boundary function call is made. The compiler process allocates a single global offset table pointer and then pre-appends a word to each function entry point containing the offset from that function entry point to the global offset table pointer. A calling module accesses the global offset table pointer and loads a global offset table register based on the offset and function entry point. Pointer logic instructions are inserted before pointer based branches using instruction cycles that would otherwise be lost to a pipeline stall in the preferred architecture.