Processors
Basilisk has no processor support of its own to write or maintain. It reads the processor descriptions Ghidra ships, unchanged, so the list of processors is Ghidra’s list.
Language ids
Section titled “Language ids”A language is named by four parts separated by colons, exactly as in Ghidra:
x86:LE:64:default │ │ │ └── variant │ │ └────── address size in bits │ └───────── byte order: LE little-endian, BE big-endian └───────────── processorSeveral ids can share one SLEIGH specification and differ only in their processor specification or defaults; that is why there are more ids (193) than specifications (150).
The modules
Section titled “The modules”Counts are for the Ghidra commit Basilisk is developed against.
| Module | Language ids | Specifications | An id from it |
|---|---|---|---|
| 6502 | 2 | 2 | 6502:LE:16:default |
| 68000 | 5 | 5 | 68000:BE:32:default |
| 8048 | 1 | 1 | 8048:LE:16:default |
| 8051 | 5 | 5 | 8051:BE:16:default |
| 8085 | 1 | 1 | 8085:LE:16:default |
| AARCH64 | 5 | 3 | AARCH64:LE:64:v8A |
| ARM | 22 | 16 | ARM:LE:32:v8 |
| Atmel (AVR, AVR32) | 5 | 5 | avr32:BE:32:default |
| BPF | 1 | 1 | BPF:LE:32:default |
| CP1600 | 1 | 1 | CP1600:BE:16:default |
| CR16 | 1 | 2 | CR16C:LE:16:default |
| Dalvik | 13 | 11 | Dalvik:LE:32:default |
| DATA | 2 | 2 | DATA:LE:64:default |
| eBPF | 2 | 2 | eBPF:BE:64:default |
| HCS08 | 6 | 3 | HC05:BE:16:default |
| HCS12 | 4 | 3 | HC-12:BE:16:default |
| Hexagon | 1 | 1 | Hexagon:LE:32:default |
| JVM | 1 | 1 | JVM:BE:32:default |
| Loongarch | 4 | 4 | Loongarch:LE:32:ilp32f |
| M16C | 2 | 2 | M16C/60:LE:16:default |
| M8C | 1 | 1 | M8C:BE:16:default |
| MC6800 | 3 | 3 | 6809:BE:16:default |
| MCS96 | 1 | 1 | MCS96:LE:16:default |
| MIPS | 22 | 6 | MIPS:BE:32:default |
| NDS32 | 2 | 2 | NDS32:BE:32:default |
| PA-RISC | 1 | 1 | pa-risc:BE:32:default |
| PIC | 13 | 13 | PIC-24E:LE:24:default |
| PowerPC | 23 | 18 | PowerPC:BE:32:default |
| RISCV | 3 | 3 | RISCV:LE:64:default |
| Sparc | 2 | 2 | sparc:BE:32:default |
| SuperH | 3 | 3 | SuperH:BE:32:SH-2A |
| SuperH4 | 2 | 2 | SuperH4:BE:32:default |
| TI_MSP430 | 2 | 2 | TI_MSP430:LE:16:default |
| Toy (Ghidra’s test languages) | 13 | 13 | Toy:BE:32:default |
| tricore | 4 | 1 | tricore:LE:32:default |
| V850 | 2 | 2 | V850:LE:32:default |
| x86 | 6 | 2 | x86:LE:64:default |
| Xtensa | 2 | 2 | Xtensa:LE:32:default |
| Z80 | 4 | 2 | z80:LE:16:default |
The ids of a module are in its .ldefs file, under
Ghidra/Processors/<module>/data/languages/.
Your own processor
Section titled “Your own processor”Because the files are read as they are, a SLEIGH specification written for Ghidra works in Basilisk: put the
module’s directory beside the others under Ghidra/Processors/, compile it with
basilisk sleigh compile, and its ids are available to
basilisk sleigh decode.
What each file is for
Section titled “What each file is for”| File | What it holds | Read by Basilisk |
|---|---|---|
.slaspec, .sinc |
The instruction set in SLEIGH: encodings, display, p-code | Yes; compiled |
.sla |
The compiled form of a .slaspec |
Yes; also written, byte for byte |
.ldefs |
The module’s languages: ids, descriptions, which files belong to each | Yes |
.pspec |
Processor details: the program counter, default context, register groups, default symbols | Yes |
.cspec |
A compiler’s conventions: calling conventions, stack, data sizes, p-code snippets | Yes |
.opinion |
Which language and compiler a loader should pick for a file | Yes |
.dwarf, .gdis, patterns, manuals’ indexes |
Debug-register maps, external disassembler setup, function-start patterns | With the features that use them (M5, M6) |
Planned
Section titled “Planned”| What | Milestone |
|---|---|
| Registers as a model, instruction flow and operand types, per-instruction overrides | M1, later item |
Choosing a language for a file from the .opinion files during import |
M4 |
| Relocation handlers per processor | M4 |
| Function-start patterns and per-processor analyzers | M5 |
| Processor emulation tests | M6 |