Add instructions cmovl, cmovle, cmovg

v420v · Jul 7, 2023 · df2ef54 · df2ef54
1 parent 5ec0a97
commit df2ef54
Show file tree

Hide file tree

Showing 6 changed files with 8,052 additions and 7,849 deletions.
diff --git a/doc/docs.md b/doc/docs.md
@@ -258,6 +258,15 @@ VAS uses AT&T assembly syntax. Some key differences between AT&T syntax and Inte
 - [x] `cmovgeq`
 - [x] `cmovgel`
 - [x] `cmovgew`
+- [x] `cmovlq`
+- [x] `cmovll`
+- [x] `cmovlw`
+- [x] `cmovleq`
+- [x] `cmovlel`
+- [x] `cmovlew`
+- [x] `cmovgq`
+- [x] `cmovgl`
+- [x] `cmovgw`
 - [x] `ret`
 - [x] `syscall`
 - [x] `nop`

diff --git a/doc/ドキュメント.md b/doc/ドキュメント.md
@@ -261,6 +261,15 @@ VAS は AT&T アセンブリ構文を使用しています。AT&T 構文と他
 - [x] `cmovgeq`
 - [x] `cmovgel`
 - [x] `cmovgew`
+- [x] `cmovlq`
+- [x] `cmovll`
+- [x] `cmovlw`
+- [x] `cmovleq`
+- [x] `cmovlel`
+- [x] `cmovlew`
+- [x] `cmovgq`
+- [x] `cmovgl`
+- [x] `cmovgw`
 - [x] `ret`
 - [x] `syscall`
 - [x] `nop`

diff --git a/encoder/addr.v b/encoder/addr.v
@@ -49,7 +49,7 @@ fn (mut e Encoder) change_symbol_visibility(instr Instr, visibility u8) {
 fn (mut e Encoder) fix_same_section_relocations() {
 	for mut rela in e.rela_text_users {
 		if symbol := e.user_defined_symbols[rela.uses] {
-			if symbol.section != rela.instr.section {
+			if symbol.section_name != rela.instr.section_name {
 				continue
 			}
 			if symbol.binding == encoder.stb_global {
@@ -64,10 +64,15 @@ fn (mut e Encoder) fix_same_section_relocations() {
 
 			mut hex := [u8(0), 0, 0, 0]
 			binary.little_endian_put_u32(mut &hex, u32(num))
-			e.user_defined_sections[rela.instr.section].code[rela.instr.addr + rela.offset] = hex[0]
-			e.user_defined_sections[rela.instr.section].code[rela.instr.addr + rela.offset+1] = hex[1]
-			e.user_defined_sections[rela.instr.section].code[rela.instr.addr + rela.offset+2] = hex[2]
-			e.user_defined_sections[rela.instr.section].code[rela.instr.addr + rela.offset+3] = hex[3]
+
+			mut section := e.user_defined_sections[rela.instr.section_name] or {
+				panic('this should not happen')
+			}
+
+			section.code[rela.instr.addr + rela.offset] = hex[0]
+			section.code[rela.instr.addr + rela.offset+1] = hex[1]
+			section.code[rela.instr.addr + rela.offset+2] = hex[2]
+			section.code[rela.instr.addr + rela.offset+3] = hex[3]
 
 			rela.is_already_resolved = true
 		}
@@ -76,30 +81,36 @@ fn (mut e Encoder) fix_same_section_relocations() {
 
 pub fn (mut e Encoder) assign_addresses() {
 	for mut instr in e.instrs {
-		if instr.section !in e.user_defined_sections {
-			e.user_defined_sections[instr.section] = &UserDefinedSection{}
+		if instr.kind == .section && instr.section_name !in e.user_defined_sections {
+			e.user_defined_sections[instr.section_name] = &UserDefinedSection{
+				flags: section_flags(instr.flags)
+			}
 		}
-		mut section := e.user_defined_sections[instr.section] or {
+
+		mut section := e.user_defined_sections[instr.section_name] or {
 			panic('this should not happen')
 		}
+
 		match instr.kind {
-			.section {
-				section.flags = section_flags(instr.flags)
-			}
 			.global {
 				e.change_symbol_binding(*instr, encoder.stb_global)
+				continue
 			}
 			.local {
 				e.change_symbol_binding(*instr, encoder.stb_local)
+				continue
 			}
 			.hidden {
 				e.change_symbol_visibility(*instr, encoder.stv_hidden)
+				continue
 			}
 			.internal {
 				e.change_symbol_visibility(*instr, encoder.stv_internal)
+				continue
 			}
 			.protected {
 				e.change_symbol_visibility(*instr, encoder.stv_protected)
+				continue
 			} else {}
 		}
 

diff --git a/encoder/directive.v b/encoder/directive.v
@@ -4,19 +4,19 @@ import token
 import error
 import encoding.binary
 
-fn (mut e Encoder) add_section(name string, flag string, pos token.Position) {
-	e.current_section = name
+fn (mut e Encoder) add_section(section_name string, flag string, pos token.Position) {
+	e.current_section_name = section_name
 
-	instr := Instr{kind: .section, pos: pos, section: name, symbol_type: encoder.stt_section, flags: flag}
+	instr := Instr{kind: .section, pos: pos, section_name: section_name, symbol_type: encoder.stt_section, flags: flag}
 	e.instrs << &instr
 
-	if s := e.user_defined_symbols[name] {
+	if s := e.user_defined_symbols[section_name] {
 		if s.kind == .label {
-			error.print(pos, 'symbol `$name` is already defined')
+			error.print(pos, 'symbol `$section_name` is already defined')
 			exit(1)
 		}
 	} else {
-		e.user_defined_symbols[name] = &instr
+		e.user_defined_symbols[section_name] = &instr
 	}
 }
 

diff --git a/encoder/encoder.v b/encoder/encoder.v
@@ -10,9 +10,9 @@ mut:
 	tok token.Token // current token
 	l   lexer.Lexer // lexer
 pub mut:
-	current_section       string
+	current_section_name  string
 	current_instr         &Instr
-	instrs                []&Instr
+	instrs                []&Instr // All instructions, sections, symbols, directives
 	rela_text_users       []Rela
 	user_defined_symbols  map[string]&Instr
 	user_defined_sections map[string]&UserDefinedSection
@@ -100,7 +100,10 @@ pub enum InstrKind {
 	leave
 	cmovs
 	cmovns
+	cmovg
 	cmovge
+	cmovl
+	cmovle
 	cvttss2sil
 	cvtsi2ssq
 	cvtsi2sdq
@@ -198,7 +201,7 @@ pub mut:
 	binding        u8
 	visibility     u8 // STV_DEFAULT, STV_INTERNAL, STV_HIDDEN, STV_PROTECTED
 	symbol_type    u8
-	section        string         [required]
+	section_name   string         [required]
 	is_jmp_or_call bool
 	pos            token.Position [required]
 }
@@ -331,7 +334,7 @@ pub fn new(mut l lexer.Lexer, file_name string) &Encoder {
 	mut e := &Encoder{
 		tok: tok
 		l: l
-		current_section: '.text'
+		current_section_name: '.text'
 		instrs: []&Instr{cap: 1500000}
 		current_instr: unsafe { nil }
 	}
@@ -345,7 +348,7 @@ fn (mut e Encoder) set_current_instr(kind InstrKind) {
 	instr := &Instr{
 		pos: e.tok.pos
 		kind: kind
-		section: e.current_section
+		section_name: e.current_section_name
 	}
 	e.current_instr = instr
 	e.instrs << instr
@@ -651,7 +654,7 @@ fn (mut e Encoder) encode_instr() {
 		instr := Instr{
 			kind: .label
 			pos: pos
-			section: e.current_section
+			section_name: e.current_section_name
 			symbol_name: instr_name
 		}
 		e.expect(.colon)
@@ -683,7 +686,7 @@ fn (mut e Encoder) encode_instr() {
 			e.instrs << &Instr{
 				kind: .global
 				pos: pos
-				section: e.current_section
+				section_name: e.current_section_name
 				symbol_name: e.tok.lit
 			}
 			e.next()
@@ -692,7 +695,7 @@ fn (mut e Encoder) encode_instr() {
 			e.instrs << &Instr{
 				kind: .local
 				pos: pos
-				section: e.current_section
+				section_name: e.current_section_name
 				symbol_name: e.tok.lit
 			}
 			e.next()
@@ -701,7 +704,7 @@ fn (mut e Encoder) encode_instr() {
 			e.instrs << &Instr{
 				kind: .hidden
 				pos: pos
-				section: e.current_section
+				section_name: e.current_section_name
 				symbol_name: e.tok.lit
 			}
 			e.next()
@@ -710,7 +713,7 @@ fn (mut e Encoder) encode_instr() {
 			e.instrs << &Instr{
 				kind: .internal
 				pos: pos
-				section: e.current_section
+				section_name: e.current_section_name
 				symbol_name: e.tok.lit
 			}
 			e.next()
@@ -719,7 +722,7 @@ fn (mut e Encoder) encode_instr() {
 			e.instrs << &Instr{
 				kind: .protected
 				pos: pos
-				section: e.current_section
+				section_name: e.current_section_name
 				symbol_name: e.tok.lit
 			}
 			e.next()
@@ -1021,78 +1024,87 @@ fn (mut e Encoder) encode_instr() {
 		'CMOVNSQ', 'CMOVNSL', 'CMOVNSW' {
 			e.cmov(.cmovns, [u8(0x0F), 0x49], get_size_by_suffix(instr_name_upper))
 		}
+		'CMOVLQ', 'CMOVLL', 'CMOVLW' {
+			e.cmov(.cmovl, [u8(0x0F), 0x4C], get_size_by_suffix(instr_name_upper))
+		}
 		'CMOVGEQ', 'CMOVGEL', 'CMOVGEW' {
 			e.cmov(.cmovge, [u8(0x0F), 0x4D], get_size_by_suffix(instr_name_upper))
 		}
+		'CMOVLEQ', 'CMOVLEL', 'CMOVLEW' {
+			e.cmov(.cmovle, [u8(0x0F), 0x4E], get_size_by_suffix(instr_name_upper))
+		}
+		'CMOVGQ', 'CMOVGL', 'CMOVGW' {
+			e.cmov(.cmovg, [u8(0x0F), 0x4F], get_size_by_suffix(instr_name_upper))
+		}
 		'RETQ', 'RET' {
 			e.instrs << &Instr{
 				kind: .ret
 				pos: pos
-				section: e.current_section
+				section_name: e.current_section_name
 				code: [u8(0xc3)]
 			}
 		}
 		'SYSCALL' {
 			e.instrs << &Instr{
 				kind: .syscall
 				pos: pos
-				section: e.current_section
+				section_name: e.current_section_name
 				code: [u8(0x0f), 0x05]
 			}
 		}
 		'NOPQ', 'NOP' {
 			e.instrs << &Instr{
 				kind: .nop
 				pos: pos
-				section: e.current_section
+				section_name: e.current_section_name
 				code: [u8(0x90)]
 			}
 		}
 		'HLT' {
 			e.instrs << &Instr{
 				kind: .hlt
 				pos: pos
-				section: e.current_section
+				section_name: e.current_section_name
 				code: [u8(0xf4)]
 			}
 		}
 		'LEAVE' {
 			e.instrs << &Instr{
 				kind: .leave
 				pos: pos
-				section: e.current_section
+				section_name: e.current_section_name
 				code: [u8(0xc9)]
 			}
 		}
 		'CLTQ' {
 			e.instrs << &Instr{
 				kind: .cltq
 				pos: pos
-				section: e.current_section
+				section_name: e.current_section_name
 				code: [u8(0x48), 0x98]
 			}
 		}
 		'CLTD' {
 			e.instrs << &Instr{
 				kind: .cltd
 				pos: pos
-				section: e.current_section
+				section_name: e.current_section_name
 				code: [u8(0x99)]
 			}
 		}
 		'CQTO' {
 			e.instrs << &Instr{
 				kind: .cqto
 				pos: pos
-				section: e.current_section
+				section_name: e.current_section_name
 				code: [u8(0x48), 0x99]
 			}
 		}
 		'CWTL' {
 			e.instrs << &Instr{
 				kind: .cwtl
 				pos: pos
-				section: e.current_section
+				section_name: e.current_section_name
 				code: [u8(0x98)]
 			}
 		}