native: implement for-c and for-in range loops (#12155)

cmd/tools/vtest-all.v

@@ -71,7 +71,13 @@ fn get_all_commands() []Command {
 			line: '$vexe run examples/v_script.vsh > /dev/null'
 			okmsg: 'V can run the .VSH script file examples/v_script.vsh'
 		}
-		//
+		$if linux {
+			res << Command{
+				line: '$vexe -b native run examples/native/hello_world.v > /dev/null'
+				okmsg: 'V compiles and runs examples/native/hello_world.v on the native backend for linux'
+			}
+		}
+		// only compilation:
 		res << Command{
 			line: '$vexe -os linux -b native -o hw.linux examples/hello_world.v'
 			okmsg: 'V compiles hello_world.v on the native backend for linux'

examples/native/hello_world.v

@@ -1,20 +1,18 @@
-// fn println(s string) { }
-
-// fn test_fn() {
-// println('test fn')
-//}
+fn test_fn() {
+	println('test fn')
+}
 
 fn main() {
 	println('native test')
-	// i := 0
-	// for i < 5 {
-	for _ in 1 .. 5 {
-		println('Hello world from V native machine code generator!')
-		// i++
+	mut i := 0
+	for i < 5 {
+		println('Hello world from V native machine code generator, in a while style for loop!')
+		i++
+	}
+	for _ in 1 .. 5 {
+		println('Hello world from V native machine code generator, in a range loop!')
 	}
-	/*
 	println('Hello again!')
-	//test_fn()
+	test_fn()
 	println('done')
-	*/
 }

vlib/v/gen/native/amd64.v

@@ -105,18 +105,36 @@ fn (mut g Gen) cmp(reg Register, size Size, val i64) {
 	g.println('cmp $reg, $val')
 }
 
-/*
-rax // 0
-	rcx // 1
-	rdx // 2
-	rbx // 3
-	rsp // 4
-	rbp // 5
-	rsi // 6
-	rdi // 7
-*/
-// `a == 1`
-// `cmp DWORD [rbp-0x4],0x1`
+// `cmp rax, rbx`
+fn (mut g Gen) cmp_reg(reg Register, reg2 Register) {
+	match reg {
+		.rax {
+			match reg2 {
+				.rbx {
+					g.write([byte(0x48), 0x39, 0xd8])
+				}
+				else {
+					g.n_error('Cannot compare $reg and $reg2')
+				}
+			}
+		}
+		.rbx {
+			match reg2 {
+				.rax {
+					g.write([byte(0x48), 0x39, 0xc3])
+				}
+				else {
+					g.n_error('Cannot compare $reg and $reg2')
+				}
+			}
+		}
+		else {
+			g.n_error('Cannot compare $reg and $reg2')
+		}
+	}
+	g.println('cmp $reg, $reg2')
+}
+
 fn (mut g Gen) cmp_var(var_name string, val int) {
 	g.write8(0x81) // 83 for 1 byte?
 	g.write8(0x7d)
@@ -126,6 +144,15 @@ fn (mut g Gen) cmp_var(var_name string, val int) {
 	g.println('cmp var `$var_name` $val')
 }
 
+// `sub DWORD [rbp-0x4], 1`
+fn (mut g Gen) dec_var(var_name string) {
+	g.write16(0x6d81) // 83 for 1 byte
+	offset := g.get_var_offset(var_name)
+	g.write8(0xff - offset + 1)
+	g.write32(1)
+	g.println('dec_var `$var_name`')
+}
+
 // `add DWORD [rbp-0x4], 1`
 fn (mut g Gen) inc_var(var_name string) {
 	g.write16(0x4581) // 83 for 1 byte
@@ -260,7 +287,7 @@ fn (mut g Gen) mov_reg_to_var(var_offset int, reg Register) {
 fn (mut g Gen) mov_var_to_reg(reg Register, var_offset int) {
 	// 8b 7d f8    mov edi,DWORD PTR [rbp-0x8]
 	match reg {
-		.rax, .rsi {
+		.rax, .rbx, .rsi {
 			g.write8(0x48)
 		}
 		else {}
@@ -616,6 +643,7 @@ fn (mut g Gen) lea(reg Register, val int) {
 	g.write8(0x8d)
 	g.write8(0x15)
 	g.write32(val)
+	g.println('lea $reg, $val')
 }
 
 fn (mut g Gen) mov(reg Register, val int) {
@@ -626,7 +654,7 @@ fn (mut g Gen) mov(reg Register, val int) {
 				g.write8(0xc7)
 				g.write8(0xc0)
 				g.write32(-1)
-				return
+				g.println('mov $reg, $val')
 			}
 			.rcx {
 				if val == -1 {
@@ -638,12 +666,14 @@ fn (mut g Gen) mov(reg Register, val int) {
 					g.write8(0xff)
 					g.write8(0xff) // mov rcx 0xffff5
 				}
-				return
+				g.println('mov $reg, $val')
 			}
 			else {
 				g.n_error('unhandled mov $reg, -1')
 			}
 		}
+		g.println('mov $reg, $val')
+		return
 	}
 	if val == 0 {
 		// Optimise to xor reg, reg when val is 0
@@ -885,8 +915,7 @@ pub fn (mut g Gen) call_fn(node ast.CallExpr) {
 	} else {
 		g.call(int(addr))
 	}
-	g.println('fn call `${name}()`')
-	// println('call $name $addr')
+	g.println('call `${name}()`')
 }
 
 fn (mut g Gen) patch_calls() {
@@ -953,17 +982,35 @@ fn (mut g Gen) assign_stmt(node ast.AssignStmt) {
 						g.mov_reg_to_var(dest, .rax)
 					}
 					.decl_assign {
-						g.allocate_var(name, 4, right.val.int())
+						g.allocate_var(name, 8, right.val.int())
 					}
 					.assign {
-						match node.left_types[i] {
-							7 { // ast.IndexExpr {
+						// dump(g.typ(node.left_types[i]))
+						match node.left[i] {
+							ast.Ident {
+								// lname := '${node.left[i]}'
+								// g.expr(node.right[i])
+								g.mov(.rax, right.val.int())
+								offset := g.get_var_offset('i') // node.left[i])
+								g.mov_reg_to_var(offset, .rax)
+							}
+							ast.InfixExpr {
+								eprintln('assign')
+								// dump(node.left[i])
+								offset := g.get_var_offset('i') // node.left[i])
+								g.mov_reg_to_var(offset, native.fn_arg_registers[i])
+							}
+							/*
+							ast.int_type_idx {
+								g.expr(node.left[i])
+match node.left[i] {
+ast.IndexExpr {
 								ie := node.left[i] as ast.IndexExpr
 								bracket := name.index('[') or {
 									g.v_error('bracket expected', node.pos)
 									exit(1)
 								}
-								var_name := name[0..bracket]
+								var_name := name[0 .. bracket]
 								mut dest := g.get_var_offset(var_name)
 								index := ie.index as ast.IntegerLiteral
 								dest += index.val.int() * 8
@@ -971,7 +1018,13 @@ fn (mut g Gen) assign_stmt(node ast.AssignStmt) {
 								g.mov(.rax, right.val.int())
 								g.mov_reg_to_var(dest, .rax)
 								// eprintln('${var_name}[$index] = ${right.val.int()}')
+} else {
+dump(node)
+g.v_error('oops', node.pos)
+}
+}
 							}
+							*/
 							else {
 								tn := node.left[i].type_name()
 								dump(node.left_types)
@@ -988,12 +1041,12 @@ fn (mut g Gen) assign_stmt(node ast.AssignStmt) {
 			ast.InfixExpr {
 				// eprintln('infix') dump(node) dump(right)
 				g.infix_expr(right)
-				offset := g.allocate_var(name, 4, 0)
+				offset := g.allocate_var(name, 8, 0)
 				// `mov DWORD PTR [rbp-0x8],eax`
 				if g.pref.is_verbose {
 					println('infix assignment $name offset=$offset.hex2()')
 				}
-				g.mov_reg_to_var(offset, .eax)
+				g.mov_reg_to_var(offset, .rax)
 			}
 			ast.Ident {
 				// eprintln('identr') dump(node) dump(right)
@@ -1020,7 +1073,7 @@ fn (mut g Gen) assign_stmt(node ast.AssignStmt) {
 						g.mov_reg_to_var(dest, .rax)
 					}
 					.decl_assign {
-						dest := g.allocate_var(name, 4, 0)
+						dest := g.allocate_var(name, 8, 0)
 						g.mov_var_to_reg(.rax, g.get_var_offset(right.name))
 						g.mov_reg_to_var(dest, .rax)
 					}
@@ -1069,7 +1122,7 @@ fn (mut g Gen) assign_stmt(node ast.AssignStmt) {
 			}
 			ast.IndexExpr {
 				// a := arr[0]
-				offset := g.allocate_var(name, 4, 0)
+				offset := g.allocate_var(name, 8, 0)
 				if g.pref.is_verbose {
 					println('infix assignment $name offset=$offset.hex2()')
 				}
@@ -1131,9 +1184,6 @@ fn (mut g Gen) assign_stmt(node ast.AssignStmt) {
 }
 
 fn (mut g Gen) infix_expr(node ast.InfixExpr) {
-	if g.pref.is_verbose {
-		println('infix expr op=$node.op')
-	}
 	// TODO
 	if node.left is ast.InfixExpr {
 		g.n_error('only simple expressions are supported right now (not more than 2 operands)')
@@ -1229,31 +1279,39 @@ fn (mut g Gen) gen_asm_stmt_amd64(asm_node ast.AsmStmt) {
 		match t.name {
 			'nop' {
 				g.write8(byte(0x90))
+				g.println('nop')
 			}
 			'syscall' {
 				g.write8(byte(0x0f))
 				g.write8(byte(0x05))
+				g.println('syscall')
 			}
 			'ret' {
 				g.write8(byte(0xc3))
+				g.println('ret')
 			}
 			'int3' {
 				g.write8(byte(0xcc))
 				g.write8(byte(imm))
+				g.println('int3')
 			}
 			'sti' {
 				g.write8(byte(0xfb))
+				g.println('sti')
 			}
 			'cli' {
 				g.write8(byte(0xfa))
+				g.println('cli')
 			}
 			'int' {
 				g.write8(byte(0xcd))
 				g.write8(byte(imm))
+				g.println('int')
 			}
 			'cpuid' {
 				g.write8(byte(0x0f))
 				g.write8(byte(0xa2))
+				g.println('cpuid')
 			}
 			'mov' {
 				g.write8(byte(0xb8 + reg))
@@ -1261,6 +1319,7 @@ fn (mut g Gen) gen_asm_stmt_amd64(asm_node ast.AsmStmt) {
 				g.write8(byt(imm, 1))
 				g.write8(byt(imm, 2))
 				g.write8(byt(imm, 3))
+				g.println('mov $reg, $imm')
 			}
 			else {
 				g.v_error('unsupported instruction $t.name', asm_node.pos)
@@ -1437,8 +1496,18 @@ fn (mut g Gen) for_stmt(node ast.ForStmt) {
 		ast.Ident {
 			lit := infix_expr.right as ast.IntegerLiteral
 			g.cmp_var(infix_expr.left.name, lit.val.int())
+			match infix_expr.left.tok_kind {
+				.lt {
 					jump_addr = g.cjmp(.jge)
 				}
+				.gt {
+					jump_addr = g.cjmp(.jle)
+				}
+				else {
+					g.n_error('unhandled infix cond token')
+				}
+			}
+		}
 		else {
 			g.n_error('unhandled infix.left')
 		}

vlib/v/gen/native/gen.v

@@ -446,25 +446,84 @@ fn (mut g Gen) println(comment string) {
 	println(' ' + comment)
 }
 
-fn (mut g Gen) for_in_stmt(node ast.ForInStmt) {
-	g.v_error('for-in statement is not yet implemented', node.pos)
-	/*
-	if node.is_range {
-		// `for x in 1..10 {`
-		// i := if node.val_var == '_' { g.new_tmp_var() } else { c_name(node.val_var) }
-		// val_typ := g.table.mktyp(node.val_type)
-		g.write32(0x3131) // 'for (${g.typ(val_typ)} $i = ')
+fn (mut g Gen) gen_forc_stmt(node ast.ForCStmt) {
+	if node.has_init {
+		g.stmts([node.init])
+	}
+	start := g.pos()
+	mut jump_addr := i64(0)
+	if node.has_cond {
+		cond := node.cond
+		match cond {
+			ast.InfixExpr {
+				// g.infix_expr(node.cond)
+				match mut cond.left {
+					ast.Ident {
+						lit := cond.right as ast.IntegerLiteral
+						g.cmp_var(cond.left.name, lit.val.int())
+						match cond.op {
+							.gt {
+								jump_addr = g.cjmp(.jle)
+							}
+							.lt {
+								jump_addr = g.cjmp(.jge)
+							}
+							else {
+								g.n_error('unsupported conditional in for-c loop')
+							}
+						}
+					}
+					else {
+						g.n_error('unhandled infix.left')
+					}
+				}
+			}
+			else {}
+		}
+		// dump(node.cond)
 		g.expr(node.cond)
-		g.write32(0x3232) // ; $i < ')
-		g.expr(node.high)
-		g.write32(0x3333) // '; ++$i) {')
+	}
+	g.stmts(node.stmts)
+	if node.has_inc {
+		g.stmts([node.inc])
+	}
+	g.jmp(int(0xffffffff - (g.pos() + 5 - start) + 1))
+	g.write32_at(jump_addr, int(g.pos() - jump_addr - 4))
+
+	// loop back
+}
+
+fn (mut g Gen) for_in_stmt(node ast.ForInStmt) {
+	if node.stmts.len == 0 {
+		// if no statements, just dont make it
+		return
+	}
+	if node.is_range {
+		// for a in node.cond .. node.high {
+		i := g.allocate_var(node.val_var, 8, 0) // iterator variable
+		g.expr(node.cond)
+		g.mov_reg_to_var(i, .rax) // i = node.cond // initial value
+		start := g.pos() // label-begin:
+		g.mov_var_to_reg(.rbx, i) // rbx = iterator value
+		g.expr(node.high) // final value
+		g.cmp_reg(.rbx, .rax) // rbx = iterator, rax = max value
+		jump_addr := g.cjmp(.jge) // leave loop if i is beyond end
+		g.stmts(node.stmts)
+		g.inc_var(node.val_var)
+		g.jmp(int(0xffffffff - (g.pos() + 5 - start) + 1))
+		g.write32_at(jump_addr, int(g.pos() - jump_addr - 4))
+		/*
 		} else if node.kind == .array {
 	} else if node.kind == .array_fixed {
 	} else if node.kind == .map {
 	} else if node.kind == .string {
 	} else if node.kind == .struct_ {
-	}
+	} else if it.kind in [.array, .string] || it.cond_type.has_flag(.variadic) {
+	} else if it.kind == .map {
 		*/
+	} else {
+		g.v_error('for-in statement is not yet implemented', node.pos)
+	}
 }
 
 pub fn (mut g Gen) gen_exit(node ast.Expr) {
@@ -487,6 +546,9 @@ fn (mut g Gen) stmt(node ast.Stmt) {
 		ast.FnDecl {
 			g.fn_decl(node)
 		}
+		ast.ForCStmt {
+			g.gen_forc_stmt(node)
+		}
 		ast.ForInStmt {
 			g.for_in_stmt(node)
 		}
@@ -619,7 +681,10 @@ fn (mut g Gen) expr(node ast.Expr) {
 		ast.ArrayInit {
 			g.n_error('array init expr not supported yet')
 		}
-		ast.BoolLiteral {}
+		ast.BoolLiteral {
+			g.mov64(.rax, if node.val { 1 } else { 0 })
+			eprintln('bool literal')
+		}
 		ast.CallExpr {
 			if node.name == 'C.syscall' {
 				g.gen_syscall(node)
@@ -633,7 +698,12 @@ fn (mut g Gen) expr(node ast.Expr) {
 			}
 		}
 		ast.FloatLiteral {}
-		ast.Ident {}
+		ast.Ident {
+			offset := g.get_var_offset(node.obj.name) // i := 0
+			// offset := g.get_var_offset(node.name)
+			// XXX this is intel specific
+			g.mov_var_to_reg(.rax, offset)
+		}
 		ast.IfExpr {
 			if node.is_comptime {
 				eprintln('Warning: ignored compile time conditional not yet supported for the native backend.')
@@ -641,8 +711,15 @@ fn (mut g Gen) expr(node ast.Expr) {
 				g.if_expr(node)
 			}
 		}
-		ast.InfixExpr {}
-		ast.IntegerLiteral {}
+		ast.InfixExpr {
+			g.infix_expr(node)
+			// get variable by name
+			// save the result in rax
+		}
+		ast.IntegerLiteral {
+			g.mov64(.rax, node.val.int())
+			// g.gen_print_reg(.rax, 3, fd)
+		}
 		ast.PostfixExpr {
 			g.postfix_expr(node)
 		}
@@ -669,9 +746,15 @@ fn (mut g Gen) postfix_expr(node ast.PostfixExpr) {
 	}
 	ident := node.expr as ast.Ident
 	var_name := ident.name
-	if node.op == .inc {
+	match node.op {
+		.inc {
 			g.inc_var(var_name)
 		}
+		.dec {
+			g.dec_var(var_name)
+		}
+		else {}
+	}
 }
 
 [noreturn]

vlib/v/gen/native/tests/forin.vv

@@ -0,0 +1,26 @@
+fn main() {
+	mut i := 0
+	for i = 0; i < 3; i++ {
+		println('loop0')
+	}
+	i = 0
+	for i < 3 {
+		println('loop1')
+		i++
+	}
+	for _ in 0 .. 3 {
+		println('loop2')
+	}
+	n := 3
+	for _ in 0 .. n {
+		println('loop3')
+	}
+	for i = 3; i > 0; i-- {
+		println('loop4')
+	}
+	i = 3
+	for i > 0 {
+		println('loop5')
+		i--
+	}
+}

vlib/v/gen/native/tests/forin.vv.out

@@ -0,0 +1,18 @@
+loop0
+loop0
+loop0
+loop1
+loop1
+loop1
+loop2
+loop2
+loop2
+loop3
+loop3
+loop3
+loop4
+loop4
+loop4
+loop5
+loop5
+loop5