This post provides my look on MP ASM and an attempt to analyze and decompose it on parts. Despite this post provides ‘partitioning’ of a MP ASM only, many similarities can be found in other assemblers.
Although macro-assembler in most cases is referred as a single language, it worth to mention that it actually consists of two languages – ASM part and macro part. This is especially noticeable when switching from one family to another – macro part remains the same while ASM part is noticeably different due to difference in the instruction sets. Besides these two languages, there are also some other languages involved in process of developing an ASM application.
Table 1 gives list of identified programming languages (PL), their inputs and productions:
| Language | Operates on | Produces | Description |
|---|---|---|---|
| ASM | Device resources | Object files | Target device language |
| Macro | Macro variables and ASM statements | Expanded ASM – intermediate source | PL operating on ASM constructions |
| Preprocessor | Input files and source code lines | Preprocessed source | PL operating on source lines |
| LKR | Object files and sections, device definition | Executable file | PL operating on objects produced by ASM, It is also used to provide definition of device resources, so linker may properly allocate them. |
| MCP | Source files | Commands the compiler and linker | A kind of “PL” used to specify what sources/files constitute a project |
| DEV | Device definition | Device simulation model | PL for defining device architecture and resources for the simulator |
| SCL | Device and application model | Visual simulation or traces | a VHDL-like simulation language |
In some cases these languages provide overlapping capabilities, such as preprocessor’s #define vs macro’s equ and macro’s .org vs linker’s section start=.
From this set of languages only ASM is a “real” device language, whose statements will be actually executed by the device(s). The other languages are tool languages, executed by the development tools.
Also there is a pseudo language used for expressing statement semantics (A <- B – A). Such pseudo language is necessary, because the mnemonics themselves are not descriptive enough (not “human readable”). Having just a mnemonic it is hard to say what the action it implies and what objects it affects. Furthermore, similar mnemonics given by different vendors may imply different semantics, which make it even more complicated.
Post a Comment