feat(cmd): add gasm dis and share the decoder with the debugger

2026-09-14 18:47:08 +02:00
parent 50db6615b2
commit edc20ffa97
7 changed files with 405 additions and 26 deletions
@@ -0,0 +1,95 @@
+// Copyright (c) 2026 Petr Balvín <opensource@petrbalvin.org> (https://petrbalvin.org)
+// SPDX-License-Identifier: BSD-3-Clause
+
+// Package disasm decodes machine code back to instruction text for the four
+// architectures gasm assembles.  It is a thin, platform-independent wrapper
+// over golang.org/x/arch and backs both the `gasm dis` command and the live
+// debugger views.
+package disasm
+
+import (
+	"fmt"
+
+	"golang.org/x/arch/arm64/arm64asm"
+	"golang.org/x/arch/loong64/loong64asm"
+	"golang.org/x/arch/riscv64/riscv64asm"
+	"golang.org/x/arch/x86/x86asm"
+
+	"sourcedock.dev/petrbalvin/gasm-devkit/arch"
+)
+
+// Instruction is one decoded instruction: its text form, its length in bytes
+// and the address it was decoded at.
+type Instruction struct {
+	Addr uint64
+	Text string
+	Len  int
+}
+
+// Decode decodes the instruction at the start of code, located at addr.
+// code needs to hold at least the one instruction being decoded (amd64 may
+// consume up to 15 bytes).  Undecodable bytes yield the placeholder text "???"
+// and a length of one word (four bytes, one on amd64) so that a listing can
+// keep making progress, mirroring the debugger's behaviour.
+func Decode(a arch.Arch, code []byte, addr uint64) (Instruction, error) {
+	if len(code) == 0 {
+		return Instruction{}, fmt.Errorf("disasm: empty input")
+	}
+	switch a {
+	case arch.ARM64:
+		if len(code) < 4 {
+			return Instruction{}, fmt.Errorf("disasm: need 4 bytes, have %d", len(code))
+		}
+		inst, err := arm64asm.Decode(code)
+		if err != nil {
+			return Instruction{Addr: addr, Text: "???", Len: 4}, nil
+		}
+		return Instruction{Addr: addr, Text: arm64asm.GoSyntax(inst, addr, nil, nil), Len: 4}, nil
+
+	case arch.RISCV:
+		// The compressed extensions are decoded transparently; a 16-bit
+		// instruction only needs its two bytes.
+		inst, err := riscv64asm.Decode(code)
+		if err != nil {
+			return Instruction{Addr: addr, Text: "???", Len: 2}, nil
+		}
+		return Instruction{Addr: addr, Text: riscv64asm.GoSyntax(inst, addr, nil, nil), Len: inst.Len}, nil
+
+	case arch.LOONG64:
+		if len(code) < 4 {
+			return Instruction{}, fmt.Errorf("disasm: need 4 bytes, have %d", len(code))
+		}
+		inst, err := loong64asm.Decode(code)
+		if err != nil {
+			return Instruction{Addr: addr, Text: "???", Len: 4}, nil
+		}
+		return Instruction{Addr: addr, Text: loong64asm.GoSyntax(inst, addr, nil), Len: 4}, nil
+
+	default: // amd64
+		inst, err := x86asm.Decode(code, 64)
+		if err != nil {
+			return Instruction{Addr: addr, Text: "???", Len: 1}, nil
+		}
+		return Instruction{Addr: addr, Text: x86asm.IntelSyntax(inst, addr, nil), Len: inst.Len}, nil
+	}
+}
+
+// Block decodes up to max instructions from code starting at addr and returns
+// them in order.  Decoding stops at the end of code or once an instruction
+// would run past it.
+func Block(a arch.Arch, code []byte, addr uint64, max int) []Instruction {
+	var out []Instruction
+	pc := 0
+	for len(out) < max && pc < len(code) {
+		ins, err := Decode(a, code[pc:], addr+uint64(pc))
+		if err != nil {
+			break
+		}
+		if ins.Len <= 0 || pc+ins.Len > len(code) {
+			break
+		}
+		out = append(out, ins)
+		pc += ins.Len
+	}
+	return out
+}
@@ -0,0 +1,141 @@
+// Copyright (c) 2026 Petr Balvín <opensource@petrbalvin.org> (https://petrbalvin.org)
+// SPDX-License-Identifier: BSD-3-Clause
+
+package disasm
+
+import (
+	"strings"
+	"testing"
+
+	"sourcedock.dev/petrbalvin/gasm-devkit/arch"
+	"sourcedock.dev/petrbalvin/gasm-devkit/asm"
+	"sourcedock.dev/petrbalvin/gasm-devkit/ast"
+	"sourcedock.dev/petrbalvin/gasm-devkit/parser"
+)
+
+func TestDecodeKnownBytes(t *testing.T) {
+	for _, tt := range []struct {
+		a    arch.Arch
+		code []byte
+		text string
+		want int
+	}{
+		{arch.AMD64, []byte{0x55}, "push rbp", 1},
+		{arch.AMD64, []byte{0x48, 0x89, 0xE5}, "mov rbp, rsp", 3},
+		{arch.ARM64, []byte{0xc0, 0x03, 0x5f, 0xd6}, "RET", 4},
+		{arch.RISCV, []byte{0x67, 0x80, 0x00, 0x00}, "RET", 4},
+		{arch.LOONG64, []byte{0x20, 0x00, 0x00, 0x4c}, "RET", 4},
+	} {
+		ins, err := Decode(tt.a, tt.code, 0)
+		if err != nil {
+			t.Errorf("%s: %v", tt.a, err)
+			continue
+		}
+		if ins.Text != tt.text || ins.Len != tt.want {
+			t.Errorf("%s: % x decoded to %q (%d bytes), want %q (%d)",
+				tt.a, tt.code, ins.Text, ins.Len, tt.text, tt.want)
+		}
+	}
+}
+
+func TestDecodeUndecodable(t *testing.T) {
+	// Zero words do not encode a usable instruction on arm64 and loong64; the
+	// placeholder keeps a listing going.  RISC-V is the exception: an all-zero
+	// word is the defined UNIMP instruction.
+	for _, a := range []arch.Arch{arch.ARM64, arch.LOONG64} {
+		ins, err := Decode(a, []byte{0, 0, 0, 0}, 0)
+		if err != nil {
+			t.Fatalf("%s: %v", a, err)
+		}
+		if ins.Text != "???" {
+			t.Errorf("%s: text = %q, want ???", a, ins.Text)
+		}
+	}
+	// The compressed quadrant claims the zero halfword first, so the zero
+	// word decodes as the 2-byte compressed UNIMP.
+	if ins, err := Decode(arch.RISCV, []byte{0, 0, 0, 0}, 0); err != nil || ins.Text != "UNIMP" || ins.Len != 2 {
+		t.Errorf("riscv zero word: %q len %d err %v, want UNIMP with 2 bytes", ins.Text, ins.Len, err)
+	}
+	if _, err := Decode(arch.ARM64, []byte{0, 0}, 0); err == nil {
+		t.Error("short input: expected an error")
+	}
+	if _, err := Decode(arch.AMD64, nil, 0); err == nil {
+		t.Error("empty input: expected an error")
+	}
+}
+
+// TestBlockRoundTrip assembles a small kernel with the gasm encoder for every
+// architecture and disassembles it back: the listing must cover the whole
+// function and end in RET.
+func TestBlockRoundTrip(t *testing.T) {
+	for _, tt := range []struct {
+		a    arch.Arch
+		name string
+	}{
+		{arch.AMD64, "k_amd64.s"},
+		{arch.ARM64, "k_arm64.s"},
+		{arch.RISCV, "k_riscv64.s"},
+		{arch.LOONG64, "k_loong64.s"},
+	} {
+		src := "TEXT \u00b7k(SB), NOSPLIT, $0\n\tMOVQ AX, CX\n\tRET\n"
+		if tt.a != arch.AMD64 {
+			src = "TEXT \u00b7k(SB), NOSPLIT, $0\n\tRET\n"
+		}
+		f, errs := parser.Parse(tt.name, src)
+		if len(errs) > 0 {
+			t.Fatalf("%s: parse: %v", tt.a, errs)
+		}
+		img, err := assemble(t, tt.a, f)
+		if err != nil {
+			t.Fatalf("%s: assemble: %v", tt.a, err)
+		}
+		fn := img.Funcs[0]
+		code := img.Code[fn.Offset : fn.Offset+fn.Size]
+
+		ins := Block(tt.a, code, 0, 100)
+		if len(ins) == 0 {
+			t.Fatalf("%s: empty listing", tt.a)
+		}
+		consumed := 0
+		for _, in := range ins {
+			if in.Text == "" || in.Text == "???" {
+				t.Errorf("%s: undecoded instruction at %#x: %q", tt.a, in.Addr, in.Text)
+			}
+			consumed += in.Len
+		}
+		if consumed != len(code) {
+			t.Errorf("%s: listing consumed %d of %d bytes", tt.a, consumed, len(code))
+		}
+		if last := ins[len(ins)-1]; !strings.Contains(strings.ToLower(last.Text), "ret") {
+			t.Errorf("%s: last instruction = %q, want RET", tt.a, last.Text)
+		}
+	}
+}
+
+func TestBlockLimits(t *testing.T) {
+	code := []byte{0x55, 0x55, 0x55, 0x55, 0x55}
+	if got := Block(arch.AMD64, code, 0, 3); len(got) != 3 {
+		t.Errorf("max=3 produced %d instructions, want 3", len(got))
+	}
+	if got := Block(arch.AMD64, code, 0, 100); len(got) != 5 {
+		t.Errorf("code end produced %d instructions, want 5", len(got))
+	}
+	if got := Block(arch.AMD64, nil, 0, 3); len(got) != 0 {
+		t.Errorf("empty code produced %d instructions, want 0", len(got))
+	}
+}
+
+// assemble assembles the parsed file with the encoder for a.
+func assemble(t *testing.T, a arch.Arch, f *ast.File) (*asm.Image, error) {
+	t.Helper()
+	switch a {
+	case arch.ARM64:
+		return asm.AssembleFileARM64(f)
+	case arch.RISCV:
+		return asm.AssembleFileRISCV(f)
+	case arch.LOONG64:
+		return asm.AssembleFileLOONG64(f)
+	default:
+		return asm.AssembleFile(f)
+	}
+}