Relocating machine instructions by currying

• Authors: Norman Ramsey
• Venue: PLDI '96 Proceedings of the ACM SIGPLAN 1996 conference on Programming language design and implementation
• Year: 1996

Relocation adjusts machine instructions to account for changes in the locations of the instructions themselves or of external symbols to which they refer. Standard linkers implement a finite set of relocation transformations, suitable for a single architecture. These transformations are enumerated, named, and engraved in a machine-dependent object-file format, and linkers must recognize them by name. These names and their associated transformations are an unnecessary source of machine-dependence.The New Jersey Machine-Code Toolkit is an application generator. It helps programmers create applications that manipulate machine code, including linkers. Guided by a short instruction-set specification, the toolkit generates the bit-manipulating code. Instructions are described by constructors, which denote functions mapping lists of operands to instructions' binary representations. Any operand can be designated as "relocatable," meaning that the operand's value need not be known at the time the instruction is encoded. For instructions with relocatable operands, the toolkit computes relocating transformations. Tool writers can use the toolkit to create machine-independent software that relocates machine instructions. mld, a retargetable linker built with the toolkit, needs only 20 lines of C code for relocation, and that code is machine-independent.The toolkit discovers relocating transformations by currying encoding functions. An attempt to encode an instruction with a relocatable operand results in the creation of a closure. The closure can be applied when the values of the relocatable operands become known. Currying provides a general, machine-independent method of relocation.Currying rewrites a λ-term into two nested λ-terms. The standard implementation has the first λ allocate a closure and store therein its operands and a pointer to the second λ. Using this strategy in the toolkit means that, when it builds an application, the toolkit generates code for many different inner λ-terms---one for each instruction that uses a relocatable address. Hoisting some of the computation out of the second λ into the first makes many of the second λs identical---a handful are enough for a whole instruction set. This optimization reduces the size of machine-dependent assembly and linking code by 15--20% for the MIPS, SPARC, and PowerPC, and by about 30% for the Pentium. It also makes the second λs equivalent to relocating transformations named in standard object-file formats.