avoid using Zig std for builtin translation

Author: Vexu

src/Translator.zig

@@ -13,6 +13,7 @@ const Type = aro.Type;
 const ast = @import("ast.zig");
 const ZigNode = ast.Node;
 const ZigTag = ZigNode.Tag;
+const builtins = @import("builtins.zig");
 const Scope = @import("Scope.zig");
 
 const Translator = @This();
@@ -55,6 +56,9 @@ tree: Tree,
 comp: *aro.Compilation,
 mapper: aro.TypeMapper,
 
+rendered_builtins: std.EnumSet(builtins.Builtin) = .{},
+render_buf: std.ArrayList(u8),
+
 fn getMangle(t: *Translator) u32 {
     t.mangle_count += 1;
     return t.mangle_count;
@@ -137,15 +141,15 @@ pub fn translate(
     gpa: mem.Allocator,
     comp: *aro.Compilation,
     tree: aro.Tree,
-) !std.zig.Ast {
+) ![]u8 {
     const mapper = tree.comp.string_interner.getFastTypeMapper(tree.comp.gpa) catch tree.comp.string_interner.getSlowTypeMapper();
     defer mapper.deinit(tree.comp.gpa);
 
     var arena_allocator = std.heap.ArenaAllocator.init(gpa);
     defer arena_allocator.deinit();
     const arena = arena_allocator.allocator();
 
-    var context: Translator = .{
+    var translator: Translator = .{
         .gpa = gpa,
         .arena = arena,
         .alias_list = .empty,
@@ -154,39 +158,39 @@ pub fn translate(
         .comp = comp,
         .mapper = mapper,
         .tree = tree,
+        .render_buf = .init(gpa),
     };
-    context.global_scope.* = Scope.Root.init(&context);
+    translator.global_scope.* = Scope.Root.init(&translator);
     defer {
-        context.decl_table.deinit(gpa);
-        context.alias_list.deinit(gpa);
-        context.global_names.deinit(gpa);
-        context.weak_global_names.deinit(gpa);
-        context.opaque_demotes.deinit(gpa);
-        context.unnamed_typedefs.deinit(gpa);
-        context.typedefs.deinit(gpa);
-        context.global_scope.deinit();
-        context.pattern_list.deinit(gpa);
-    }
-
-    inline for (@typeInfo(std.zig.c_builtins).@"struct".decls) |decl| {
-        const builtin_fn = try ZigTag.pub_var_simple.create(arena, .{
-            .name = decl.name,
-            .init = try ZigTag.import_c_builtin.create(arena, decl.name),
-        });
-        try addTopLevelDecl(&context, decl.name, builtin_fn);
+        translator.decl_table.deinit(gpa);
+        translator.alias_list.deinit(gpa);
+        translator.global_names.deinit(gpa);
+        translator.weak_global_names.deinit(gpa);
+        translator.opaque_demotes.deinit(gpa);
+        translator.unnamed_typedefs.deinit(gpa);
+        translator.typedefs.deinit(gpa);
+        translator.global_scope.deinit();
+        translator.pattern_list.deinit(gpa);
+        translator.render_buf.deinit();
     }
 
-    try prepopulateGlobalNameTable(&context);
-    try transTopLevelDecls(&context);
+    try prepopulateGlobalNameTable(&translator);
+    try transTopLevelDecls(&translator);
 
-    for (context.alias_list.items) |alias| {
-        if (!context.global_scope.sym_table.contains(alias.alias)) {
+    for (translator.alias_list.items) |alias| {
+        if (!translator.global_scope.sym_table.contains(alias.alias)) {
             const node = try ZigTag.alias.create(arena, .{ .actual = alias.alias, .mangled = alias.name });
-            try addTopLevelDecl(&context, alias.alias, node);
+            try addTopLevelDecl(&translator, alias.alias, node);
         }
     }
 
-    return ast.render(gpa, context.global_scope.nodes.items);
+    var zig_ast = try ast.render(gpa, translator.global_scope.nodes.items);
+    defer {
+        gpa.free(zig_ast.source);
+        zig_ast.deinit(gpa);
+    }
+    try zig_ast.renderToArrayList(&translator.render_buf, .{});
+    return translator.render_buf.toOwnedSlice();
 }
 
 fn prepopulateGlobalNameTable(t: *Translator) !void {
@@ -1352,6 +1356,9 @@ fn transExpr(t: *Translator, scope: *Scope, expr: NodeIndex, used: ResultUsed) T
 
         .shl_expr => return t.transShiftExpr(scope, expr, .shl, used),
         .shr_expr => return t.transShiftExpr(scope, expr, .shr, used),
+
+        .builtin_call_expr => return t.transBuiltinCall(scope, expr, used),
+        .builtin_call_expr_one => return t.transBuiltinCall(scope, expr, used),
         else => unreachable, // Not an expression.
     }
 }
@@ -1582,6 +1589,58 @@ fn transPointerArithmeticSignedOp(
     return t.transCreateNodeInfixOp(if (is_add) .add else .sub, lhs_node, bitcast_node, used);
 }
 
+fn transBuiltinCall(
+    t: *Translator,
+    scope: *Scope,
+    call_node: NodeIndex,
+    used: ResultUsed,
+) TransError!ZigNode {
+    // TODO would be nice to have a helper to extract the builtin name
+    const builtin_name, const arg_nodes = switch (t.nodeTag(call_node)) {
+        .builtin_call_expr => blk: {
+            const range = t.nodeData(call_node).range;
+            const name = t.tree.tokSlice(@intFromEnum(t.tree.data[range.start]));
+            const arg_nodes = t.tree.data[range.start + 1 .. range.end];
+            break :blk .{ name, arg_nodes };
+        },
+        .builtin_call_expr_one => blk: {
+            const decl = t.nodeData(call_node).decl;
+            const name = t.tree.tokSlice(decl.name);
+            const ptr: [*]const NodeIndex = @ptrCast(&decl.node);
+            const slice = ptr[0..1];
+            const end = std.mem.indexOfScalar(NodeIndex, slice, .none) orelse 1;
+            const arg_nodes = slice[0..end];
+            break :blk .{ name, arg_nodes };
+        },
+        else => unreachable,
+    };
+    const builtin = std.meta.stringToEnum(builtins.Builtin, builtin_name) orelse {
+        const call_loc = 0; // TODO builtin call source location
+        return t.fail(error.UnsupportedTranslation, call_loc, "TODO implement function '{s}' in std.zig.c_builtins", .{builtin_name});
+    };
+
+    if (builtin.source()) |source| {
+        if (!t.rendered_builtins.contains(builtin)) {
+            t.rendered_builtins.insert(builtin);
+            try t.render_buf.appendSlice(source);
+        }
+
+        const args = try t.arena.alloc(ZigNode, arg_nodes.len);
+        for (arg_nodes, args) |arg_node, *arg| {
+            arg.* = try t.transExpr(scope, arg_node, .used);
+        }
+
+        const res = try ZigTag.call.create(t.arena, .{
+            .lhs = try ZigTag.fn_identifier.create(t.arena, builtin_name),
+            .args = args,
+        });
+        if (t.nodeType(call_node).is(.void)) return res;
+        return t.maybeSuppressResult(used, res);
+    }
+
+    @panic("TODO transBulitinCall others");
+}
+
 // =====================
 // Node creation helpers
 // =====================

src/ast.zig

@@ -785,7 +785,7 @@ pub fn render(gpa: Allocator, nodes: []const Node) !std.zig.Ast {
     defer ctx.tokens.deinit(gpa);
 
     // Estimate that each top level node has 10 child nodes.
-    const estimated_node_count = nodes.len * 10;
+    const estimated_node_count = nodes.len * 10 + 1; // +1 for the .root node
     try ctx.nodes.ensureTotalCapacity(gpa, estimated_node_count);
     // Estimate that each each node has 2 tokens.
     const estimated_tokens_count = estimated_node_count * 2;

src/builtins.zig

@@ -0,0 +1,93 @@
+pub const Builtin = enum {
+    __builtin_bswap16,
+    __builtin_bswap32,
+    __builtin_bswap64,
+    __builtin_signbit,
+    __builtin_signbitf,
+    __builtin_popcount,
+    __builtin_ctz,
+    __builtin_clz,
+    __builtin_sqrt,
+    __builtin_sqrtf,
+    __builtin_sin,
+    __builtin_sinf,
+    __builtin_cos,
+    __builtin_cosf,
+    __builtin_exp,
+    __builtin_expf,
+    __builtin_exp2,
+    __builtin_exp2f,
+    __builtin_log,
+    __builtin_logf,
+    __builtin_log2,
+    __builtin_log2f,
+    __builtin_log10,
+    __builtin_log10f,
+    __builtin_abs,
+    __builtin_labs,
+    __builtin_llabs,
+    __builtin_fabs,
+    __builtin_fabsf,
+    __builtin_floor,
+    __builtin_floorf,
+    __builtin_ceil,
+    __builtin_ceilf,
+    __builtin_trunc,
+    __builtin_truncf,
+    __builtin_round,
+    __builtin_roundf,
+    __builtin_strlen,
+    __builtin_strcmp,
+    __builtin_object_size,
+    __builtin___memset_chk,
+    __builtin_memset,
+    __builtin___memcpy_chk,
+    __builtin_memcpy,
+    __builtin_nanf,
+    __builtin_huge_valf,
+    __builtin_inff,
+    __builtin_isnan,
+    __builtin_isinf,
+    __builtin_isinf_sign,
+    __has_builtin,
+    __builtin_assume,
+    __builtin_unreachable,
+    __builtin_constant_p,
+    __builtin_mul_overflow,
+
+    // __builtin_alloca_with_align is not currently implemented.
+    // It is used in a run and a translate test to ensure that non-implemented
+    // builtins are correctly demoted. If you implement __builtin_alloca_with_align,
+    // please update the tests to use a different non-implemented builtin.
+
+    pub fn source(b: Builtin) ?[]const u8 {
+        return switch (b) {
+            .__builtin_signbit => @embedFile("builtins/signbit.zig"),
+            .__builtin_signbitf => @embedFile("builtins/signbitf.zig"),
+            .__builtin_popcount => @embedFile("builtins/popcount.zig"),
+            .__builtin_ctz => @embedFile("builtins/ctz.zig"),
+            .__builtin_clz => @embedFile("builtins/clz.zig"),
+            .__builtin_abs => @embedFile("builtins/abs.zig"),
+            .__builtin_labs => @embedFile("builtins/labs.zig"),
+            .__builtin_llabs => @embedFile("builtins/llabs.zig"),
+            .__builtin_strlen => @embedFile("builtins/strlen.zig"),
+            .__builtin_strcmp => @embedFile("builtins/strcmp.zig"),
+            .__builtin_object_size => @embedFile("builtins/object_size.zig"),
+            .__builtin___memset_chk => @embedFile("builtins/memset_chk.zig"),
+            .__builtin_memset => @embedFile("builtins/memset.zig"),
+            .__builtin___memcpy_chk => @embedFile("builtins/memcpy_chk.zig"),
+            .__builtin_memcpy => @embedFile("builtins/memcpy.zig"),
+            .__builtin_nanf => @embedFile("builtins/nanf.zig"),
+            .__builtin_huge_valf => @embedFile("builtins/huge_valf.zig"),
+            .__builtin_inff => @embedFile("builtins/inff.zig"),
+            .__builtin_isnan => @embedFile("builtins/isnan.zig"),
+            .__builtin_isinf => @embedFile("builtins/isinf.zig"),
+            .__builtin_isinf_sign => @embedFile("builtins/isinf_sign.zig"),
+            .__has_builtin => @embedFile("builtins/has_builtin.zig"),
+            .__builtin_assume => @embedFile("builtins/assume.zig"),
+            .__builtin_constant_p => @embedFile("builtins/constant_p.zig"),
+            .__builtin_mul_overflow => @embedFile("builtins/mul_overflow.zig"),
+            else => return null,
+        };
+    }
+};

src/builtins/abs.zig

@@ -0,0 +1,4 @@
+/// Standard C Library bug: The absolute value of the most negative integer remains negative.
+pub inline fn __builtin_abs(val: c_int) c_int {
+    return if (val == @import("std").math.minInt(c_int)) val else @intCast(@abs(val));
+}

src/builtins/assume.zig

@@ -0,0 +1,3 @@
+pub inline fn __builtin_assume(cond: bool) void {
+    if (!cond) unreachable;
+}

src/builtins/clz.zig

@@ -0,0 +1,6 @@
+/// Returns the number of leading 0-bits in x, starting at the most significant bit position.
+/// In C if `val` is 0, the result is undefined; in zig it's the number of bits in a c_uint
+pub inline fn __builtin_clz(val: c_uint) c_int {
+    @setRuntimeSafety(false);
+    return @as(c_int, @bitCast(@as(c_uint, @clz(val))));
+}

src/builtins/constant_p.zig

@@ -0,0 +1,4 @@
+pub inline fn __builtin_constant_p(expr: anytype) c_int {
+    _ = expr;
+    return @intFromBool(false);
+}

src/builtins/ctz.zig

@@ -0,0 +1,6 @@
+/// Returns the number of trailing 0-bits in val, starting at the least significant bit position.
+/// In C if `val` is 0, the result is undefined; in zig it's the number of bits in a c_uint
+pub inline fn __builtin_ctz(val: c_uint) c_int {
+    @setRuntimeSafety(false);
+    return @as(c_int, @bitCast(@as(c_uint, @ctz(val))));
+}

src/builtins/has_builtin.zig

@@ -0,0 +1,4 @@
+pub inline fn __has_builtin(func: anytype) c_int {
+    _ = func;
+    return @intFromBool(true);
+}

src/builtins/huge_valf.zig

@@ -0,0 +1,3 @@
+pub inline fn __builtin_huge_valf() f32 {
+    return @import("std").math.inf(f32);
+}

src/builtins/inff.zig

@@ -0,0 +1,3 @@
+pub inline fn __builtin_inff() f32 {
+    return @import("std").math.inf(f32);
+}

src/builtins/isinf.zig

@@ -0,0 +1,3 @@
+pub inline fn __builtin_isinf(x: anytype) c_int {
+    return @intFromBool(@import("std").math.isInf(x));
+}

src/builtins/isinf_sign.zig

@@ -0,0 +1,5 @@
+/// Similar to isinf, except the return value is -1 for an argument of -Inf and 1 for an argument of +Inf.
+pub inline fn __builtin_isinf_sign(x: anytype) c_int {
+    if (!@import("std").math.isInf(x)) return 0;
+    return if (@import("std").math.isPositiveInf(x)) 1 else -1;
+}

src/builtins/isnan.zig

@@ -0,0 +1,3 @@
+pub inline fn __builtin_isnan(x: anytype) c_int {
+    return @intFromBool(@import("std").math.isNan(x));
+}

src/builtins/labs.zig

@@ -0,0 +1,4 @@
+/// Standard C Library bug: The absolute value of the most negative integer remains negative.
+pub inline fn __builtin_labs(val: c_long) c_long {
+    return if (val == @import("std").math.minInt(c_long)) val else @intCast(@abs(val));
+}

src/builtins/llabs.zig

@@ -0,0 +1,4 @@
+/// Standard C Library bug: The absolute value of the most negative integer remains negative.
+pub inline fn __builtin_llabs(val: c_longlong) c_longlong {
+    return if (val == @import("std").math.minInt(c_longlong)) val else @intCast(@abs(val));
+}

src/builtins/memcpy.zig

@@ -0,0 +1,11 @@
+pub inline fn __builtin_memcpy(
+    noalias dst: ?*anyopaque,
+    noalias src: ?*const anyopaque,
+    len: usize,
+) ?*anyopaque {
+    if (len > 0) @memcpy(
+        @as([*]u8, @ptrCast(dst.?))[0..len],
+        @as([*]const u8, @ptrCast(src.?)),
+    );
+    return dst;
+}

src/builtins/memcpy_chk.zig

@@ -0,0 +1,13 @@
+pub inline fn __builtin___memcpy_chk(
+    noalias dst: ?*anyopaque,
+    noalias src: ?*const anyopaque,
+    len: usize,
+    remaining: usize,
+) ?*anyopaque {
+    if (len > remaining) @panic("__builtin___memcpy_chk called with len > remaining");
+    if (len > 0) @memcpy(
+        @as([*]u8, @ptrCast(dst.?))[0..len],
+        @as([*]const u8, @ptrCast(src.?)),
+    );
+    return dst;
+}

src/builtins/memset.zig

@@ -0,0 +1,5 @@
+pub inline fn __builtin_memset(dst: ?*anyopaque, val: c_int, len: usize) ?*anyopaque {
+    const dst_cast = @as([*c]u8, @ptrCast(dst));
+    @memset(dst_cast[0..len], @as(u8, @bitCast(@as(i8, @truncate(val)))));
+    return dst;
+}

src/builtins/memset_chk.zig

@@ -0,0 +1,11 @@
+pub inline fn __builtin___memset_chk(
+    dst: ?*anyopaque,
+    val: c_int,
+    len: usize,
+    remaining: usize,
+) ?*anyopaque {
+    if (len > remaining) @panic("__builtin___memset_chk called with len > remaining");
+    const dst_cast = @as([*c]u8, @ptrCast(dst));
+    @memset(dst_cast[0..len], @as(u8, @bitCast(@as(i8, @truncate(val)))));
+    return dst;
+}

src/builtins/mul_overflow.zig

@@ -0,0 +1,5 @@
+pub fn __builtin_mul_overflow(a: anytype, b: anytype, result: *@TypeOf(a, b)) c_int {
+    const res = @mulWithOverflow(a, b);
+    result.* = res[0];
+    return res[1];
+}