36 lines
1.3 KiB
Markdown
36 lines
1.3 KiB
Markdown
Creole is a bytecode designed for simple implementations.
|
|
|
|
## Bytecode Format
|
|
|
|
Each creole line consists of pseudo-UTF-8 characters. The first byte
|
|
is an unsigned number between 0 and 127 (the high bit is clear). Each
|
|
suceeding pseudo-UTF-8 character is encoded as follows:
|
|
|
|
* `110HHHHx 10xxxxxx`
|
|
* `1110HHHH 10xxxxxx 10xxxxxx`
|
|
* `11110HHH 10Hxxxxx 10xxxxxx 10xxxxxx`
|
|
* `111110HH 10HHxxxx 10xxxxxx 10xxxxxx 10xxxxxx`
|
|
* `1111110H 10HHHxxx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx`
|
|
* `11111110 10HHHHxx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx`
|
|
|
|
The first four bytes determine the type. The LSB high bit determines
|
|
if the encoded value is a register (`0001`) or immediate (`0010`).
|
|
The second bit from LSB determines if the value should be treated
|
|
as a signed 32 bit two's compliment number (`001X`) or should be
|
|
treated as an unsigned 32 bit number (`000X`).
|
|
|
|
All other values are reserved. Overlong values are allowed, and for some
|
|
argument values they are necessary. All lines are terminated by a byte
|
|
of all zeros.
|
|
|
|
## Assembler
|
|
|
|
The macro assembler is Python (see the asm directory).
|
|
|
|
## Design Philsophy
|
|
|
|
Creole is small but not minimal. It is easy to add instructions, and
|
|
the amount of labels, registers, and stack space can changed at runtime.
|
|
System calls can be added dynamically, but static instructions that take
|
|
at most 3 arguments should be hard-coded.
|