The goal of this project is to have a generic assembler available for assembling arbitrary instruction sets into machine code suitable for use in Logisim.
Currently, work is being done to transform smasm.py into a more readable, and generic assembler, logiasm.py. As of this commit, logiasm is not functional. See the todo list below for progress.
Argument parsing- Logging infrastructure
- ISA spec definition
Filesystem semantics- Syntax
- Parser
- Runtime error checking
- 1st pass overhaul
- Label mapping
- Pseudo-op expansion
- Recursive?
- 2nd pass overhaul
- Label resolution
- Static verification
- Register access
- Over/underflow
- Label usage
- Encoding
Below are bonus items, which will come after a working assembler
- Generic target system (Assembling for more than just Logisim)
- Test instruction generation
- Operation tagging (for generated pseudo instructions for free)
- Automated datapath debugging
logiasm is designed to be a generic assembler, which should be able to take any RISC architecture, and output machine code suitable for direct loading into Logisim. Obviously there are limitations to what "any" means, but if you can express your target ISA within the confines of the ISA file syntax, then logiasm should be able to handle it.
To define your own ISA, you'll need to create a file my_isa_name.isa in the isa directory. This enables the assembler to automatically load your ISA when it is the target of an assembly file. Otherwise, you must specify the name of your ISA file at runtime with the --isa flag.
All ISA files are specified in JSON, and must conform to the following spec. Note that order does not matter, and the bold keys are required.
registers : [int | object]
This should either specify the number of usable registers as an int, or pass an object with a key for each register mapping to an object describing the register traits. There are no required keys in the register trait object, but optional values include a boolean write key telling the assembler if writes to this register are allowed in user code, a , and a
bits: : int
The bit length of each instruction.
instructions : object
An object with instruction names for keys mapping to object values. The value object conforms to the following structure:
- type : string - If applicable, the type of instruction. Note that every instruction must supply a type, an opcode, or both
- opcode : string - A string representation of the binary opcode. The string can be optionally prefixed with "0b"
- fnct : string - The function code, generally used by instructions that share the same opcode
types : object
SMIPS (Simple MIPS) was the instruction set for which LogiAsm was originally written, then called SMasm. This version is preserved under the legacy directory.