REL32

Rel32 is a relocation type used in 32-bit x86 object formats to represent a 32-bit signed displacement that is relative to the program counter. It is commonly found in formats such as ELF for Linux and COFF/PE for Windows, where the relocation type may be referred to as REL32 or IMAGE_REL_BASED_REL32, depending on the format.

Purpose and concept

Rel32 enables position-independent addressing by encoding a relative offset rather than an absolute address. The 32-bit value at the relocation site is adjusted so that it equals the signed displacement from the address of the instruction following the relocation (the relocation site plus 4 bytes) to the target symbol. This makes the code and data references relocatable when the program is loaded at a different base address.

How it works

During linking or loading, the linker patches the relocation site with a 32-bit signed displacement. The resulting value typically represents target_address − (relocation_address + 4). The exact formula can vary slightly by format and addends, but the core idea is a PC-relative, 32-bit displacement used by instructions such as calls, jumps, or data references that rely on relative addressing.

Usage and scope

Rel32 is primarily used for PC-relative addressing within 32-bit x86 binaries. It supports references to symbols within the same binary or across linked objects, as long as the final displacement fits within a signed 32-bit range (approximately ±2 GB). It is a key mechanism for creating position-independent code in 32-bit environments.

Limitations and modern context

Because it relies on a 32-bit relative offset, very large address spaces or cross-module references exceeding the 32-bit range cannot use REL32. In modern 64-bit code, RIP-relative addressing is preferred, and 32-bit REL32-like relocations are less common.