rune

A DSL compiler that generates instruction encoders for any architecture. Designed and written for rasi

What

A description of instruction encodings are written in a .rn file, Rune will read and emit a generated assembler.hpp which is pure inline. This can be dropped straight into your project.

DSL

(inst mov_ri64 (enc 0xB8+rd /imm64)                   (pattern (mov (reg64 $rd) (imm64 $imm))))
(inst mov_rr64 (enc 0x89 ModRM(r $rs rm $rd))         (pattern (mov (reg64 $rd) (reg64 $rs))))
(inst add_rr64 (enc 0x01 ModRM(r $rs rm $rd))         (pattern (add (reg64 $rd) (reg64 $rs))))
(inst sub_rr64 (enc 0x01 ModRM(r $rs rm $rd))         (pattern (sub (reg64 $rd) (reg64 $rs))))
(inst ret      (enc 0xC3)                             (pattern (ret)))

Encoding Atoms

Atom	Description
0xNN	Literal byte
0xNN+rd	Opcode byte with lower 3 bits of register encoded in
ModRM(r $x rm $y)	ModRM byte from two operands
/imm8, /imm32, /imm64	Emit immediate value of given width
/disp8, /disp32	Emit displacement
/0 ... /7	ModRM opcode extension

REX prefixes are derived from operand types automatically, reg64 and imm64 impl REX.W. Additionally, extended registers which is r8 to r15 set REX.R/B automatically.

Usage

rune x86_64.rn -o assembler.hpp --ns rasi::x86_64 --prefix emit_

CLI Flags

Flag	Default	Description
-o <file>	assembler.hpp	Output file
--ns <ns>	x86_64	C++ namespace
--prefix <p>	emit_	Function name prefix
--case <lower/upper>	lower	Instruction name casing