Toy project allowing to compile and run (on x86 machines) programs that are written in custom assembly of 'W-machine' computer (example implementation) - simplified computer model, designed in Computer Science Department of Silesian University of Technology for teaching basics of Computer Architecture.
W-machine is based on von Neumann architecture.
All operations are performed using single accumulator register, for which in this project x86 register %eax is used as the virtual 32-bit ACC.
gcc wmjit.c -m32 -I utilis tokenizer.c compiler.c utilis/utilis.c utilis/strlist.c utilis/strbuilder.c utilis/exmem.c -o bin/wmjit
Supports Windows and Linux systems (see exmem.c file).
Regarding 64-bit systems:
- On Windows machines this should work out of the box for
x86_64GCC. - On Linux, one might be required to install
gcc-multilibto supprot building 32-bit application. Kernels like Ubuntu are compiled withCONFIG_IA32_EMULATION=yflags, so we are able to run 32-bit programs directly on 64-bit system.
Program accepts input assembly code only from .prg files (see examples).
When running on linux, make sure that input file lines are not terminated with CRLF.
Following options are avaliable:
| Argument | Description |
|---|---|
-b BINFILE |
Write memory dump of .prg to BINFILE after it is compiled. |
-d BINFILE |
Write memory dump of .prg to BINFILE after it finishes running. |
--out=ADDR:FILE,... |
Set output device of ADDR to specific file (max 8 devs). |
--in=ADDR:FILE,... |
Set input device of ADDR to specific file (max 8 devs). |
--nr |
Do not run, only parse and compile. |
--help |
Prints this message. |
| Instrucion | Description |
|---|---|
NOP |
No operation |
STP |
Stop execution |
ADD [address] |
Add value at memory address to accumulator |
SUB [address] |
Subtract value at memory address from accumulator |
LOAD [address] |
Load value from memory into accumulator |
STOR [address] |
Store value from accumulator into memory address |
JMP [address] |
Unconditional jump to address |
BLZ [address] |
Branch if ACC less than zero |
BEZ [address] |
Branch if ACC equal to zero |
BGZ [address] |
Branch if ACC greater than zero |
PSH |
Push an accumulator onto stack |
POP |
Pop value from stack into accumulator |
DIV [address] |
Divide accumulator by memory value |
MUL [address] |
Multiply accumulator by memory value |
IORD [device] |
Read value from input device into accumulator |
IOWR [device] |
Write character from accumulator into output device |
DEF [value] |
Define value data word |
RES |
Reserve data word |
I/O operations using IORD and IOWR instructions are implemented as a call to specific function:
// Note: if 'device' < 0, 'c' is printed to |device| as integer not a character
__attribute__((__cdecl__))
static void __iowr(WORD_t device, int c);
// Uses fgetc to read from mapped 'device' - potential EOF should be handled within jitted code
__attribute__((__cdecl__))
static int __iord(WORD_t device);
Following device mapping is created by default (can be overwritten using --out= and --in= options):
| TYPE | DEVICE | FILE |
|---|---|---|
| OUT | 1 | stdout |
| OUT | 2 | stderr |
| IN | 1 | stdin |