Rewriting symbol address initialization sequences

1. A computer-implemented method for linking object code comprising:

recognizing a first symbol address initialization sequence in a module by determining that first symbol address initialization sequence corresponds with a pre-defined symbol address instruction sequence, wherein the first symbol address initialization sequence includes one or more instructions for loading an address in a first register of a processor, the one or more instructions including an instruction that depends on a value stored in a second register of the processor, the first register for storing an address of a table of contents (TOC) and the second register for storing a starting address of the module;

determining whether to replace the first symbol address initialization sequence with a second symbol address initialization sequence by determining whether a first condition is satisfied, wherein the first condition is that the second symbol address initialization sequence has a second processor resource usage that is less than a first processor resource usage associated with the first symbol address initialization sequence; and

replacing the first symbol address initialization sequence with the second symbol address instruction sequence when it is determined that the first condition is satisfied.

2. The method of claim 1 , wherein the first symbol address initialization sequence is position independent code.

3. The method of claim 1 , further comprising generating a procedure linkage table (PLT) stub for a function that calls the module, the PLT stub including an instruction to load the starting address of the module in the second register.

4. The method of claim 1 , wherein the recognizing of the first symbol address initialization sequence in the module includes determining that the first symbol address initialization sequence is at a beginning of a function having at least a global function entry point and a local function entry point, wherein the global entry point is an entry point of a function call when the function call is a direct and external function call or an indirect function call, and wherein the local entry point is an entry point of a function call when the function call is a direct and local function call or a call not requiring a TOC pointer value.

5. The method of claim 4 , the first condition includes that the first symbol address initialization sequence between the global and local function entry point cannot be branched into.

6. The method of claim 1 , wherein the module includes position independent code (PIC), the first symbol address initialization sequence is configured for use with position independent code (PIC), and the second symbol address initialization sequence is configured for use with position dependent code (PDC).

7. The method of claim 6 , wherein the first condition is that an address of the TOC pointer of the module is within a first or last 2 GB of a 64-bit address space.

8. The method of claim 1 , wherein the module includes position dependent code (PDC), the first symbol address initialization sequence is configured for use with position dependent code (PDC), and the second symbol address initialization sequence is configured for use with position dependent code (PDC).

9. The method of claim 8 , wherein the first condition includes that the first symbol address initialization sequence is configured to load a 64-bit address and that an address of the TOC pointer of the module is within a first or last 2 GB of a 64-bit address space.

10. The method of claim 1 , wherein the first condition includes that one of relocation information or a linker switch indicates that no branches are allowed to target an address between a first and second entry point.

11. The method of claim 1 , further comprising:

determining whether the first symbol address initialization sequence has a different number of instructions than the second symbol address initialization sequence; and

updating at least one function entry address when the first symbol address initialization sequence has a different number of instructions than the second symbol address initialization sequence.

12. The method of claim 1 , wherein the second symbol address initialization sequence includes one or more instructions for loading an address in the first register, the one or more instructions including an instruction that includes a symbol defined by a linker that refers to the address of the TOC.

13. A computer-implemented method for linking object code comprising:

recognizing a first symbol address initialization sequence in a module by determining that first symbol address initialization sequence corresponds with a pre-defined symbol address instruction sequence, wherein the first symbol address initialization sequence includes one or more instructions for loading an address in a first register of a processor, the one or more instructions including an instruction that depends on a value stored in a second register of the processor, the first register for storing an address of a table of contents (TOC) and the second register for storing a starting address of the module;

determining whether to replace the first symbol address initialization sequence with a second symbol address initialization sequence by determining whether a first condition is satisfied, wherein the first condition is that one of relocation information or a linker switch indicates that no branches are allowed to target an address between a first and second entry point; and

replacing the first symbol address initialization sequence with the second symbol address instruction sequence when it is determined that the first condition is satisfied.