Abstract: Method and apparatus for selective post-compile conversion of data protocol-specific source code elements to restore the intended functionality of the source code without requiring precompilation alteration of the latter. Byte-ordering dependent source code issues may be addressed using the current invention, possibly requiring the alteration, deletion and/or addition of one or more instructions to the post-compile output. Invention may enable portability of software programs from a first processor platform and OS to a second environment such as a second hardware and OS or a simulated, emulated or virtualized environment representing a second hardware architecture, by compiling the software program source code according to the second environment and selectively converting the data protocol specific code elements in the object/executable file, produced by the aforementioned compilation, to restore their intended functionality.

Abstract: Every function that is called (“called function”) is expanded by insertion of several statements at the entry and exit thereof. Moreover, a calling function may also be expanded, by insertion of statements prior to and/or subsequent to a statement in which a called function is invoked. Many of the statements that are inserted contain new variables (called “synthetic variables”) to which registers are allocated during register allocation; the synthetic variables are not part of the originally-written software but are introduced during expansion, e.g. to transfer arguments to and return value from the called functions. Statements that are inserted can be either statements that are translated into assembly code in the normal manner, or alternatively dummy statements that contain instructions which are never translated into assembly code (i.e. ignored by the assembler. Use of dummy statements ensures that a web is established for new variables. After expansion, register allocation is performed.

Abstract: Colors to be used in register allocation are grouped into a number of sequences. Each sequence is associated with an attribute (e.g. size and/or type) of variables whose nodes in an interference graph can be colored by colors in the sequence. In certain embodiments, in addition to the above-described grouping, colors within a group are ordered in a sequence. The specific order that is used may depend on, for example, an attribute (such as size) and a predetermined preference. One example of such a predetermined preference is that a color that represents a register of the size that is associated with the sequence is located at the front of the sequence. Another color located later in the sequence represents a register of a different size than the size associated with the sequence.

Abstract: Colors to be used in register allocation are grouped into a number of sequences. Each sequence is associated with an attribute (e.g. size and/or type) of variables whose nodes in an interference graph can be colored by colors in the sequence. In certain embodiments, in addition to the above-described grouping, colors within a group are ordered in a sequence. The specific order that is used may depend on, for example, an attribute (such as size) and a predetermined preference. One example of such a predetermined preference is that a color that represents a register of the size that is associated with the sequence is located at the front of the sequence. Another color located later in the sequence represents a register of a different size than the size associated with the sequence.