Odin/core/rexcode/mips/tablegen/gen.odin

// rexcode  ·  Brendan Punsky (dotbmp@github), original author

package rexcode_mips_tablegen

// =============================================================================
// MIPS TABLE GENERATOR  (Stage A)
// =============================================================================
//
// Reads the single-source-of-truth ENCODING_TABLE (encoding_table.odin, this
// package) and emits human-readable, type-checked Odin into ./generated/:
//
//   generated/encode_tables.odin   ENCODE_FORMS + ENCODE_RUNS (flattened encode)
//   generated/decode_tables.odin   DECODE_ENTRIES + primary/SPECIAL/REGIMM/COP1/
//                                   SPECIAL2/SPECIAL3 index tables
//   generated/writer.odin          Stage B: serialize those globals to ../tables/*.bin
//
// It also re-emits the library loader ../tables.odin. Run:
//   odin run mips/tablegen            # Stage A
//   odin run mips/tablegen/generated  # Stage B

import "core:fmt"
import "core:os"
import "core:strings"
import "core:slice"
import "core:reflect"
import "core:math/bits"
import lib "../"

// Package-scope aliases so the moved SoT resolves Mnemonic/Encoding unqualified.
Encoding :: lib.Encoding
Mnemonic :: lib.Mnemonic

Blob :: struct { global, file, typ: string }
BLOBS := [?]Blob{
	{"ENCODE_FORMS",          "mips.encode_forms.bin", "Encoding"},
	{"ENCODE_RUNS",           "mips.encode_runs.bin",  "Encode_Run"},
	{"DECODE_ENTRIES",        "mips.entries.bin",      "Decode_Entry"},
	{"DECODE_INDEX_PRIMARY",  "mips.idx_primary.bin",  "Decode_Index"},
	{"DECODE_INDEX_SPECIAL",  "mips.idx_special.bin",  "Decode_Index"},
	{"DECODE_INDEX_REGIMM",   "mips.idx_regimm.bin",   "Decode_Index"},
	{"DECODE_INDEX_COP1",     "mips.idx_cop1.bin",     "Decode_Index"},
	{"DECODE_INDEX_SPECIAL2", "mips.idx_special2.bin", "Decode_Index"},
	{"DECODE_INDEX_SPECIAL3", "mips.idx_special3.bin", "Decode_Index"},
}

DIR_GEN     :: #directory + "/generated/"
PATH_LOADER :: #directory + "/../tables.odin"

Entry :: struct {
	mnemonic:   lib.Mnemonic,
	ops:        [4]lib.Operand_Type,
	enc:        [4]lib.Operand_Encoding,
	bits:       u32,
	mask:       u32,
	feature:    lib.Feature,
	flags:      lib.Encoding_Flags,
	primary_op: u8,
	sub_key:    u8,
}

Range :: struct { start: u16, count: u16 }

main :: proc() {
	n := emit_encode_tables()
	ne := emit_decode_tables()
	emit_writer()
	emit_loader()
	fmt.printfln("mips tablegen: %d encode forms, %d decode entries", n, ne)
}

// -----------------------------------------------------------------------------
// Encode side
// -----------------------------------------------------------------------------

emit_encode_tables :: proc() -> (total: int) {
	sb := strings.builder_make()
	strings.write_string(&sb, "package rexcode_mips_generated\n\n")
	strings.write_string(&sb, "// GENERATED by ../gen.odin -- DO NOT EDIT.\n")
	strings.write_string(&sb, "// Flattened encode forms + per-mnemonic run index (source: ENCODING_TABLE).\n\n")
	strings.write_string(&sb, "import lib \"../..\"\n\n")

	for m in Mnemonic { total += len(ENCODING_TABLE[m]) }

	fmt.sbprintfln(&sb, "ENCODE_FORMS := [%d]lib.Encoding{{", total)
	for m in Mnemonic {
		forms := ENCODING_TABLE[m]
		if len(forms) == 0 { continue }
		fmt.sbprintfln(&sb, "\t// .%v", m)
		for f in forms {
			write_row(&sb, f.mnemonic, f.ops, f.enc, f.bits, f.mask, f.feature, f.flags)
		}
	}
	strings.write_string(&sb, "}\n\n")

	run_w := 0
	for m in Mnemonic { run_w = max(run_w, len(reflect.enum_string(m))) }
	strings.write_string(&sb, "ENCODE_RUNS := [lib.Mnemonic]lib.Encode_Run{\n")
	start := 0
	for m in Mnemonic {
		c := len(ENCODING_TABLE[m])
		name := reflect.enum_string(m)
		fmt.sbprintf(&sb, "\t.%s", name)
		for _ in 0..<run_w-len(name) { strings.write_byte(&sb, ' ') }
		fmt.sbprintfln(&sb, " = {{% 5d, % 3d}},", start, c)
		start += c
	}
	strings.write_string(&sb, "}\n")
	emit_file(DIR_GEN + "encode_tables.odin", &sb)
	return
}

// -----------------------------------------------------------------------------
// Decode side
// -----------------------------------------------------------------------------

emit_decode_tables :: proc() -> (total: int) {
	all: [dynamic]Entry
	defer delete(all)
	for m in Mnemonic {
		for f in ENCODING_TABLE[m] {
			primary := u8((f.bits >> 26) & 0x3F)
			sub: u8
			switch primary {
			case 0x00, 0x1C, 0x1F: sub = u8(f.bits & 0x3F)
			case 0x01:             sub = u8((f.bits >> 16) & 0x1F)
			case 0x11:             sub = u8((f.bits >> 21) & 0x1F)
			}
			append(&all, Entry{f.mnemonic, f.ops, f.enc, f.bits, f.mask, f.feature, f.flags, primary, sub})
		}
	}
	slice.sort_by(all[:], proc(a, b: Entry) -> bool {
		if a.primary_op != b.primary_op { return a.primary_op < b.primary_op }
		if a.sub_key    != b.sub_key    { return a.sub_key    < b.sub_key    }
		ac := bits.count_ones(a.mask); bc := bits.count_ones(b.mask)
		if ac != bc { return ac > bc }
		return u16(a.mnemonic) < u16(b.mnemonic)
	})

	primary_idx:  [64]Range
	special_idx:  [64]Range
	regimm_idx:   [32]Range
	cop1_idx:     [32]Range
	special2_idx: [64]Range
	special3_idx: [64]Range
	for e, i in all {
		push(&primary_idx[e.primary_op], u16(i))
		switch e.primary_op {
		case 0x00: push(&special_idx [e.sub_key], u16(i))
		case 0x01: push(&regimm_idx  [e.sub_key], u16(i))
		case 0x11: push(&cop1_idx    [e.sub_key], u16(i))
		case 0x1C: push(&special2_idx[e.sub_key], u16(i))
		case 0x1F: push(&special3_idx[e.sub_key], u16(i))
		}
	}

	sb := strings.builder_make()
	strings.write_string(&sb, "package rexcode_mips_generated\n\n")
	strings.write_string(&sb, "// GENERATED by ../gen.odin -- DO NOT EDIT.\n")
	strings.write_string(&sb, "// Reverse decode tables (source: ENCODING_TABLE), keyed by primary opcode + sub-field.\n\n")
	strings.write_string(&sb, "import lib \"../..\"\n\n")

	fmt.sbprintfln(&sb, "DECODE_ENTRIES := [%d]lib.Decode_Entry{{", len(all))
	for e in all {
		write_row(&sb, e.mnemonic, e.ops, e.enc, e.bits, e.mask, e.feature, e.flags)
	}
	strings.write_string(&sb, "}\n\n")

	emit_range(&sb, "DECODE_INDEX_PRIMARY",  primary_idx[:])
	emit_range(&sb, "DECODE_INDEX_SPECIAL",  special_idx[:])
	emit_range(&sb, "DECODE_INDEX_REGIMM",   regimm_idx[:])
	emit_range(&sb, "DECODE_INDEX_COP1",     cop1_idx[:])
	emit_range(&sb, "DECODE_INDEX_SPECIAL2", special2_idx[:])
	emit_range(&sb, "DECODE_INDEX_SPECIAL3", special3_idx[:])
	emit_file(DIR_GEN + "decode_tables.odin", &sb)
	return len(all)
}

push :: proc(r: ^Range, i: u16) { if r.count == 0 { r.start = i }; r.count += 1 }

emit_range :: proc(sb: ^strings.Builder, name: string, ranges: []Range) {
	fmt.sbprintfln(sb, "%s := [%d]lib.Decode_Index{{", name, len(ranges))
	for r, i in ranges {
		if r.count != 0 {
			fmt.sbprintfln(sb, "\t0x%02X = {{% 4d, % 3d}},", i, r.start, r.count)
		}
	}
	strings.write_string(sb, "}\n\n")
}

// -----------------------------------------------------------------------------
// Shared row + flags formatting (compact, matching mips' original generator)
// -----------------------------------------------------------------------------

write_row :: proc(sb: ^strings.Builder, mn: lib.Mnemonic, ops: [4]lib.Operand_Type,
                  enc: [4]lib.Operand_Encoding, bits, mask: u32, feature: lib.Feature, flags: lib.Encoding_Flags) {
	fmt.sbprintf(sb, "\t{{ .%v, {{.%v,.%v,.%v,.%v}}, {{.%v,.%v,.%v,.%v}}, 0x%08X, 0x%08X, .%v, {{%s}} }},\n",
		mn, ops[0], ops[1], ops[2], ops[3], enc[0], enc[1], enc[2], enc[3], bits, mask, feature, flags_lit(flags))
}

flags_lit :: proc(f: lib.Encoding_Flags) -> string {
	parts: [dynamic]string
	defer delete(parts)
	if f.delay_slot  { append(&parts, "delay_slot=true")  }
	if f.likely      { append(&parts, "likely=true")      }
	if f.only_64     { append(&parts, "only_64=true")     }
	if f.writes_hilo { append(&parts, "writes_hilo=true") }
	if f.compact     { append(&parts, "compact=true")     }
	return strings.join(parts[:], ", ", context.temp_allocator)
}

// -----------------------------------------------------------------------------
// Stage B writer + the library loader
// -----------------------------------------------------------------------------

emit_writer :: proc() {
	sb := strings.builder_make()
	strings.write_string(&sb, "package rexcode_mips_generated\n\n")
	strings.write_string(&sb, "// GENERATED by ../gen.odin -- DO NOT EDIT.\n")
	strings.write_string(&sb, "// Stage B: serialize the typed tables above to raw blobs under ../../tables/.\n\n")
	strings.write_string(&sb, "import \"core:os\"\nimport \"core:fmt\"\n\n")
	strings.write_string(&sb, "TABLES :: #directory + \"/../../tables/\"\n\n")
	strings.write_string(&sb, "raw :: #force_inline proc \"contextless\" (p: rawptr, n: int) -> []u8 {\n\treturn (cast([^]u8)p)[:n]\n}\n\n")
	strings.write_string(&sb, "w :: proc(file: string, data: []u8) {\n")
	strings.write_string(&sb, "\tif err := os.write_entire_file(file, data); err != nil {\n")
	strings.write_string(&sb, "\t\tfmt.eprintfln(\"rexcode tablegen: failed to write %s: %v\", file, err)\n\t\tos.exit(1)\n\t}\n}\n\n")
	strings.write_string(&sb, "main :: proc() {\n")
	for b in BLOBS {
		fmt.sbprintfln(&sb, "\tw(TABLES + \"%s\", raw(&%s, size_of(%s)))", b.file, b.global, b.global)
	}
	strings.write_string(&sb, "}\n")
	emit_file(DIR_GEN + "writer.odin", &sb)
}

LOADER_TYPES :: `// -----------------------------------------------------------------------------
// Subsidiary table types (generated scaffolding)
// -----------------------------------------------------------------------------

// Companion run index: ENCODE_RUNS[mnemonic] -> contiguous run in ENCODE_FORMS.
Encode_Run :: struct {
	start: u32,
	count: u32,
}

Decode_Entry :: struct #packed {
	mnemonic: Mnemonic,            // 2
	ops:      [4]Operand_Type,     // 4
	enc:      [4]Operand_Encoding, // 4
	bits:     u32,                 // 4
	mask:     u32,                 // 4
	feature:  Feature,             // 1
	flags:    Encoding_Flags,      // 1
}
#assert(size_of(Decode_Entry) == 20)

Decode_Index :: struct #packed {
	start: u16,
	count: u16,
}
#assert(size_of(Decode_Index) == 4)
`

LOADER_ACCESSORS :: `// -----------------------------------------------------------------------------
// Accessors
// -----------------------------------------------------------------------------

// Per-mnemonic encode forms: the run of ENCODE_FORMS belonging to ` + "`m`" + `.
// Replaces the old ENCODING_TABLE[m] slice; the returned view is into rodata.
@(private, require_results)
encoding_forms :: #force_inline proc "contextless" (m: Mnemonic) -> []Encoding {
	r := ENCODE_RUNS[u16(m)]
	return ENCODE_FORMS[r.start:][:r.count]
}
`

emit_loader :: proc() {
	sb := strings.builder_make()
	strings.write_string(&sb, "package rexcode_mips\n\n")
	strings.write_string(&sb, "// =============================================================================\n")
	strings.write_string(&sb, "// GENERATED FILE - DO NOT EDIT\n")
	strings.write_string(&sb, "// =============================================================================\n")
	strings.write_string(&sb, "//\n")
	strings.write_string(&sb, "// Loads the flat binary encode/decode tables into @(rodata). Produced by tablegen:\n")
	strings.write_string(&sb, "//\n")
	strings.write_string(&sb, "//   odin run tablegen            # Stage A: ENCODING_TABLE -> generated/ + this file\n")
	strings.write_string(&sb, "//   odin run tablegen/generated  # Stage B: typed Odin literals -> tables/*.bin\n")
	strings.write_string(&sb, "//\n")
	strings.write_string(&sb, "// The .bin blobs are raw, host-endian, packed struct images.\n\n")
	strings.write_string(&sb, LOADER_TYPES)
	strings.write_string(&sb, "\n// -----------------------------------------------------------------------------\n")
	strings.write_string(&sb, "// Loaded tables (rodata, embedded from tables/*.bin at compile time)\n")
	strings.write_string(&sb, "// -----------------------------------------------------------------------------\n\n")

	gmax, fmax := 0, 0
	for b in BLOBS { gmax = max(gmax, len(b.global)); fmax = max(fmax, len(b.file)) }
	for b in BLOBS {
		fmt.sbprintf(&sb, "@(rodata) %s", b.global)
		for _ in 0..<gmax-len(b.global) { strings.write_byte(&sb, ' ') }
		path := fmt.tprintf("\"tables/%s\",", b.file)
		fmt.sbprintf(&sb, " := #load(%s", path)
		for _ in 0..<fmax-len(b.file) { strings.write_byte(&sb, ' ') }
		fmt.sbprintfln(&sb, " []%s)", b.typ)
	}
	strings.write_string(&sb, "\n")
	strings.write_string(&sb, LOADER_ACCESSORS)
	emit_file(PATH_LOADER, &sb)
}

GEN_ATTRIB :: "// rexcode  ·  Brendan Punsky (dotbmp@github), original author\n\n"

emit_file :: proc(path: string, sb: ^strings.Builder) {
	if err := os.write_entire_file(path, transmute([]u8)strings.concatenate({GEN_ATTRIB, strings.to_string(sb^)})); err != nil {
		fmt.eprintfln("rexcode tablegen: failed to write %s: %v", path, err)
		os.exit(1)
	}
}