Commit f102fe93319b8fcc9b88bdef0fa69bda967d15a7 | ix

author	chloe caruso <git@paperclover.net> 2024-05-20 13:45:39-07:00
committer	chloe caruso <git@paperclover.net> 2024-05-20 13:46:36-07:00
log	f102fe93319b8fcc9b88bdef0fa69bda967d15a7
tree	b7ed26e4570086a94fb865e70f76176583a46e54
parent	cf749893b592c342cd21fe524c7c45456876cae3

perf: use more buffered writers

9 files changed, 175 insertions(+), 29 deletions(-)

CHANGELOG.md
build.zig
demo.txt
src
- compiler

CHANGELOG.md+1-1

...	...	@@ -1,6 +1,6 @@
1	1	## 0.1.1-alpha
2	2
3		- introduce `Report` and show errors during Transform.
	3	- introduce `Report` and show errors during Tokenize, Parse, and Transform.
4	4	- add flag --dump-ir for enabling ir dumps
5	5
6	6	## 0.1.0-alpha

build.zig+12

...	...	@@ -6,12 +6,23 @@ pub fn build(b: *std.Build) void {
6	6	const target = b.standardTargetOptions(.{});
7	7	const optimize = b.standardOptimizeOption(.{});
8	8
	9	const tracing = b.option(enum {
	10	none,
	11	spall,
	12	chrome,
	13	log,
	14	}, "trace", "Enable tracing") orelse .none;
	15
9	16	const dep_opts = .{ .target = target, .optimize = optimize };
10	17
11	18	const zg = b.dependency("zg", dep_opts);
12	19	const tracer = b.dependency("tracer", dep_opts);
13	20	// const ada = b.dependency("ada", dep_opts);
14	21
	22	const opts = b.addOptions();
	23	opts.addOption([]const u8, "tracing", @tagName(tracing));
	24	const opts_mod = opts.createModule();
	25
15	26	const exe = b.addExecutable(.{
16	27	.name = "compiler",
17	28	.root_source_file = b.path("src/compiler/main.zig"),
...	...	@@ -20,6 +31,7 @@ pub fn build(b: *std.Build) void {
20	31	});
21	32	b.installArtifact(exe);
22	33
	34	exe.root_module.addImport("build_options", opts_mod);
23	35	exe.root_module.addImport("zg:PropsData", zg.module("PropsData"));
24	36	exe.root_module.addImport("tracer", tracer.module("tracer"));
25	37	// if (!target.result.isWasm()) {

demo.txt+25-1

...	...	@@ -1 +1,25 @@
1		no "waay
	1	let print = @js("console").log;
	2
	3	fn doThing(x: float, y: float) -> float {
	4	let math = if x < y {
	5	x - y
	6	} else b: {
	7	let sum = x + y;
	8	if sum > 100 {
	9	break :b sum;
	10	}
	11	if x == y {
	12	return 5;
	13	}
	14	sum * -sum
	15	};
	16	if x == y {
	17	return 5;
	18	}
	19	let divisor = if x > 10 { 2 } else { 3 };
	20	math / divisor
	21	}
	22
	23	print(doThing(1, 6));
	24	print(doThing(6, 4));
	25	print(doThing(100, 52));

src/compiler/JsLink.zig+37

...	...	@@ -19,6 +19,8 @@ pub fn linkJsCode(
19	19	decls: []const Decl.Index,
20	20	writer: anytype,
21	21	) !void {
	22	var t = tracer.trace(@src());
	23	defer t.end();
22	24
23	25	// Arena allocator is used for the main flow
24	26	var link: JsLink = .{ .comp = comp };
...	...	@@ -79,6 +81,9 @@ pub fn linkJsCode(
79	81	}
80	82
81	83	fn deinit(link: *JsLink) void {
	84	var t = tracer.trace(@src());
	85	defer t.end();
	86
82	87	defer link.* = undefined;
83	88	link.decls.deinit(link.comp.gpa);
84	89	link.link_strings.deinit(link.comp.gpa);
...	...	@@ -86,6 +91,9 @@ fn deinit(link: *JsLink) void {
86	91	}
87	92
88	93	fn reserveExternSymbol(link: *JsLink, extern_symbol: String.WithLen) !void {
	94	var t = tracer.trace(@src());
	95	defer t.end();
	96
89	97	const name = extern_symbol.toSlice(&link.comp.intern_pool);
90	98	const gop = try link.link_strings.getOrPutContextAdapted(
91	99	link.comp.gpa,
...	...	@@ -104,12 +112,18 @@ fn reserveExternSymbol(link: *JsLink, extern_symbol: String.WithLen) !void {
104	112	}
105	113
106	114	fn appendLinkString(link: *JsLink, string: []const u8) !LinkString {
	115	var t = tracer.trace(@src());
	116	defer t.end();
	117
107	118	const start = link.string_bytes.items.len;
108	119	try link.string_bytes.appendSlice(link.comp.gpa, string);
109	120	return .{ .start = @intCast(start), .len = @intCast(string.len) };
110	121	}
111	122
112	123	fn declIdentifier(link: *const JsLink, decl: Decl.Index) []const u8 {
	124	var t = tracer.trace(@src());
	125	defer t.end();
	126
113	127	return (link.decls.get(decl) orelse if (std.debug.runtime_safety) {
114	128	std.debug.panic("Decl {d} ({}) is not present in link decl map", .{
115	129	@intFromEnum(decl),
...	...	@@ -119,6 +133,9 @@ fn declIdentifier(link: *const JsLink, decl: Decl.Index) []const u8 {
119	133	}
120	134
121	135	fn printDecl(link: *JsLink, decl_index: Decl.Index, writer: anytype) !void {
	136	var t = tracer.trace(@src());
	137	defer t.end();
	138
122	139	try writer.writeAll("let ");
123	140	try writer.writeAll(link.declIdentifier(decl_index));
124	141	try writer.writeAll(" = ");
...	...	@@ -137,6 +154,9 @@ fn printDecl(link: *JsLink, decl_index: Decl.Index, writer: anytype) !void {
137	154	}
138	155
139	156	fn printDeclContents(link: JsLink, tree: const JsTree, writer: anytype) !void {
	157	var t = tracer.trace(@src());
	158	defer t.end();
	159
140	160	for (tree.mainStatements()) \|stmt\| {
141	161	try link.printExpr(tree, stmt, writer);
142	162	try writer.writeAll(";\n");
...	...	@@ -144,6 +164,9 @@ fn printDeclContents(link: JsLink, tree: const JsTree, writer: anytype) !void
144	164	}
145	165
146	166	fn printBlockContents(link: JsLink, tree: const JsTree, list: JsTree.IndexList, writer: anytype) !void {
	167	var t = tracer.trace(@src());
	168	defer t.end();
	169
147	170	assert(list.slice(tree).len > 0);
148	171	for (list.slice(tree)) \|stmt\| {
149	172	try link.printExpr(tree, stmt, writer);
...	...	@@ -158,6 +181,9 @@ fn printBlockContents(link: JsLink, tree: const JsTree, list: JsTree.IndexList
158	181	}
159	182
160	183	fn printBlock(link: JsLink, tree: const JsTree, list: JsTree.IndexList, writer: anytype) !void {
	184	var t = tracer.trace(@src());
	185	defer t.end();
	186
161	187	if (list.len != 1) {
162	188	try writer.writeAll("{");
163	189	try link.printBlockContents(tree, list, writer);
...	...	@@ -168,6 +194,9 @@ fn printBlock(link: JsLink, tree: const JsTree, list: JsTree.IndexList, writer
168	194	}
169	195
170	196	fn printExpr(link: JsLink, tree: const JsTree, node: Node.Index, writer: anytype) @TypeOf(writer).Error!void {
	197	var t = tracer.trace(@src());
	198	defer t.end();
	199
171	200	switch (tree.nodeTag(node)) {
172	201	else => \|tag\| std.debug.panic("TODO: printExpr .{s}", .{@tagName(tag)}),
173	202	.root => unreachable,
...	...	@@ -328,11 +357,17 @@ fn printExpr(link: JsLink, tree: const JsTree, node: Node.Index, writer: anyty
328	357	}
329	358
330	359	fn printLocal(link: *JsLink, local: JsTree.LocalIndex, writer: anytype) !void {
	360	var t = tracer.trace(@src());
	361	defer t.end();
	362
331	363	_ = link; // todo: be smarter about this
332	364	try writer.print("v{d}", .{local});
333	365	}
334	366
335	367	fn printLabelIdentifier(link: *JsLink, local: JsTree.LocalIndex, writer: anytype) !void {
	368	var t = tracer.trace(@src());
	369	defer t.end();
	370
336	371	_ = link; // todo: be smarter about this
337	372	try writer.print("label{d}", .{local});
338	373	}
...	...	@@ -452,3 +487,5 @@ const Node = JsTree.Node;
452	487
453	488	const InternPool = @import("InternPool.zig");
454	489	const String = InternPool.String;
	490
	491	const tracer = @import("tracer");

src/compiler/JsNodeGen.zig+46

...	...	@@ -29,6 +29,9 @@ const MapEntry = union {
29	29	};
30	30
31	31	pub fn init(gpa: Allocator, arena: Allocator, ip: const InternPool, air: const Air) !JsNodeGen {
	32	const t = tracer.trace(@src());
	33	defer t.end();
	34
32	35	const air_node_map = try arena.alloc(MapEntry, air.insts.len);
33	36	if (std.debug.runtime_safety) {
34	37	@memset(air_node_map, .{ .none = {} });
...	...	@@ -52,6 +55,9 @@ pub fn init(gpa: Allocator, arena: Allocator, ip: const InternPool, air: const
52	55	}
53	56
54	57	pub fn deinit(c: *JsNodeGen) void {
	58	const t = tracer.trace(@src());
	59	defer t.end();
	60
55	61	defer c.* = undefined;
56	62	c.nodes.deinit(c.gpa);
57	63	c.extra.deinit(c.gpa);
...	...	@@ -62,6 +68,9 @@ pub fn deinit(c: *JsNodeGen) void {
62	68	}
63	69
64	70	pub fn finalize(c: *JsNodeGen) !JsTree {
	71	const t = tracer.trace(@src());
	72	defer t.end();
	73
65	74	// todo: errdefer will probably leave memory in a fully invalid state
66	75	try c.nodes.setCapacity(c.gpa, c.nodes.len);
67	76
...	...	@@ -87,6 +96,9 @@ pub fn finalize(c: *JsNodeGen) !JsTree {
87	96	}
88	97
89	98	pub fn build(c: *JsNodeGen) !void {
	99	const t = tracer.trace(@src());
	100	defer t.end();
	101
90	102	assert(c.air.instTag(.main_block) == .block);
91	103	const statements = c.air.instData(.block, .main_block).slice(c.air);
92	104
...	...	@@ -117,6 +129,9 @@ pub fn build(c: *JsNodeGen) !void {
117	129	}
118	130
119	131	fn codeGenInst(c: *JsNodeGen, inst: Air.Inst.Index) Error!void {
	132	const t = tracer.traceFmt(@src(), "#{d} {s}", .{ @intFromEnum(inst), @tagName(c.air.instTag(inst)) });
	133	defer t.end();
	134
120	135	switch (c.air.instTag(inst)) {
121	136	.arg => unreachable, // codegen must skip over these instructions.
122	137
...	...	@@ -290,6 +305,9 @@ fn codeGenInst(c: *JsNodeGen, inst: Air.Inst.Index) Error!void {
290	305	}
291	306
292	307	fn markExpression(c: *JsNodeGen, inst: Air.Inst.Index, node: Node.Index) !void {
	308	const t = tracer.trace(@src());
	309	defer t.end();
	310
293	311	const flags: Air.Flags = c.air.instFlags(inst);
294	312	const i = @intFromEnum(inst);
295	313	switch (flags.uses) {
...	...	@@ -314,6 +332,9 @@ fn markExpression(c: *JsNodeGen, inst: Air.Inst.Index, node: Node.Index) !void {
314	332	}
315	333
316	334	fn fetchExpression(c: *JsNodeGen, inst: Air.Inst.Ref) !Node.Index {
	335	const t = tracer.trace(@src());
	336	defer t.end();
	337
317	338	return switch (inst.unwrap()) {
318	339	.value => \|v\| c.valueToExpression(v),
319	340	.inst => \|index\| {
...	...	@@ -332,6 +353,9 @@ fn fetchExpression(c: *JsNodeGen, inst: Air.Inst.Ref) !Node.Index {
332	353	}
333	354
334	355	fn valueToExpression(c: *JsNodeGen, v: Value) !Node.Index {
	356	const t = tracer.trace(@src());
	357	defer t.end();
	358
335	359	const intern_tag = c.ip.items.items(.tag)[@intFromEnum(v.toIntern())];
336	360	switch (intern_tag) {
337	361	.removed => unreachable, // UAF
...	...	@@ -352,6 +376,9 @@ fn valueToExpression(c: *JsNodeGen, v: Value) !Node.Index {
352	376	}
353	377
354	378	fn addTemporary(c: *JsNodeGen) !Node.Index {
	379	const t = tracer.trace(@src());
	380	defer t.end();
	381
355	382	try c.addStmt(try c.add(.stmt_declare_local, .{
356	383	.identifier_index = c.next_id,
357	384	.init = .none,
...	...	@@ -367,6 +394,9 @@ const Branch = struct {
367	394	};
368	395
369	396	fn codeGenBranch(c: *JsNodeGen, inst: Air.Inst.Ref, temporary: ?Node.Index) !?Branch {
	397	const t = tracer.trace(@src());
	398	defer t.end();
	399
370	400	const scratch_top = c.scratch.items.len;
371	401	defer c.scratch.shrinkRetainingCapacity(scratch_top);
372	402
...	...	@@ -430,6 +460,9 @@ fn codeGenBranch(c: *JsNodeGen, inst: Air.Inst.Ref, temporary: ?Node.Index) !?Br
430	460	}
431	461
432	462	fn codeGenBlock(c: *JsNodeGen, inst: Air.Inst.Index) !void {
	463	const t = tracer.trace(@src());
	464	defer t.end();
	465
433	466	const scratch_top = c.scratch.items.len;
434	467	errdefer c.scratch.shrinkRetainingCapacity(scratch_top);
435	468
...	...	@@ -467,6 +500,9 @@ fn codeGenBlock(c: *JsNodeGen, inst: Air.Inst.Index) !void {
467	500	}
468	501
469	502	fn canLowerToExpression(c: *const JsNodeGen, inst: Air.Inst.Ref) bool {
	503	const t = tracer.trace(@src());
	504	defer t.end();
	505
470	506	return switch (inst.unwrap()) {
471	507	.value => true,
472	508	.inst => \|index\| switch (c.air.instTag(index)) {
...	...	@@ -493,6 +529,9 @@ fn canLowerToExpression(c: *const JsNodeGen, inst: Air.Inst.Ref) bool {
493	529	}
494	530
495	531	fn lowerToExpression(c: *JsNodeGen, ref: Air.Inst.Ref) !Node.Index {
	532	const t = tracer.trace(@src());
	533	defer t.end();
	534
496	535	assert(c.canLowerToExpression(ref));
497	536	return switch (ref.unwrap()) {
498	537	.value => \|v\| try c.valueToExpression(v),
...	...	@@ -504,6 +543,9 @@ fn lowerToExpression(c: *JsNodeGen, ref: Air.Inst.Ref) !Node.Index {
504	543	}
505	544
506	545	fn add(c: *JsNodeGen, comptime tag: JsTree.Tag, data: tag.Data()) !Node.Index {
	546	const t = tracer.trace(@src());
	547	defer t.end();
	548
507	549	try c.nodes.append(c.gpa, .{
508	550	.tag = tag,
509	551	.data = EncodedData.encode(tag, data),
...	...	@@ -512,6 +554,9 @@ fn add(c: *JsNodeGen, comptime tag: JsTree.Tag, data: tag.Data()) !Node.Index {
512	554	}
513	555
514	556	fn addStmt(c: *JsNodeGen, stmt: Node.Index) !void {
	557	const t = tracer.trace(@src());
	558	defer t.end();
	559
515	560	try c.scratch.append(c.arena, @intFromEnum(stmt));
516	561	}
517	562
...	...	@@ -537,3 +582,4 @@ const Air = @import("Air.zig");
537	582	const Value = @import("Value.zig");
538	583
539	584	const dod = @import("dod.zig");
	585	const tracer = @import("tracer");

src/compiler/Report.zig+16-9

...	...	@@ -183,8 +183,7 @@ pub fn deinit(report: Report, gpa: Allocator) void {
183	183	gpa.free(report.string_bytes);
184	184	}
185	185
186		pub fn dumpToWriter(report: Report, comp: *Comp, writer: anytype) !void {
187		const tty = util.detectTTYFromWriter(writer);
	186	pub fn dumpToWriter(report: Report, comp: *Comp, writer: anytype, tty: std.io.tty.Config) !void {
188	187	try writer.print("{d} errors, {d} warnings\n", .{ report.errors.len, report.warnings.len });
189	188	for (report.errors) \|item\| {
190	189	try report.formatItem(item, .@"error", comp, writer, tty);
...	...	@@ -348,13 +347,21 @@ pub fn formatItem(
348	347	}
349	348
350	349	pub fn dumpToStdErr(report: Report, comp: *Comp) !void {
351		const stderr = std.io.getStdErr().writer();
352
353		// Disabled because of coloring issues.
354		// var buf = std.io.bufferedWriter(stderr);
355		// const writer = buf.writer();
356		try dumpToWriter(report, comp, stderr);
357		// try buf.flush();
	350	const builtin = @import("builtin");
	351
	352	const stderr = std.io.getStdErr();
	353	const tty = std.io.tty.detectConfig(stderr);
	354
	355	if (builtin.os.tag == .windows and tty == .windows_api) {
	356	// Windows API uses it's own API calls, so a buffered writer
	357	// cannot be used :/
	358	try dumpToWriter(report, comp, stderr.writer(), tty);
	359	} else {
	360	var buf = std.io.bufferedWriter(stderr.writer());
	361	const writer = buf.writer();
	362	try dumpToWriter(report, comp, writer, tty);
	363	try buf.flush();
	364	}
358	365	}
359	366
360	367	pub const Builder = struct {

src/compiler/Sema.zig+5-3

...	...	@@ -92,9 +92,11 @@ pub fn deinit(s: *Sema) void {
92	92	pub fn finalize(s: *Sema) !void {
93	93	const air: Air = s.tempAir();
94	94
95		std.debug.print("=" ** 20, .{});
96		std.debug.print("\nAir for @ {d}\n", .{@intFromEnum(s.decl.teg_decl_index)});
97		air.dumpToWriter(s.ip, std.io.getStdErr().writer()) catch {};
	95	if (s.comp.options.dump_ir) {
	96	std.debug.print("=" ** 20, .{});
	97	std.debug.print("\nAir for @ {d}\n", .{@intFromEnum(s.decl.teg_decl_index)});
	98	air.dumpToWriter(s.ip, std.io.getStdErr().writer()) catch {};
	99	}
98	100
99	101	try s.comp.codeGenDecl(s.decl, &air);
100	102	}

src/compiler/Transform.zig+4-2

...	...	@@ -603,6 +603,8 @@ fn transformExpression(t: *Transform, decl: BlockContext, node: NodeIndex) Alloc
603	603	return .void_value;
604	604	};
605	605
	606	lookup.mutated = true;
	607
606	608	return switch (lookup.data) {
607	609	.declaration => try t.addInstRef(.{
608	610	.tag = .load_decl,
...	...	@@ -1235,11 +1237,11 @@ fn scratchToIndexList(t: *Transform, scratch_top: usize) !IndexList {
1235	1237	};
1236	1238	}
1237	1239
1238		fn lookupId(t: *Transform, id: []const u8) ?Scope.Symbol {
	1240	fn lookupId(t: Transform, id: []const u8) ?Scope.Symbol {
1239	1241	var i = t.scopes.items.len;
1240	1242	while (i > 0) {
1241	1243	i -= 1;
1242		if (t.scopes.items[i].symbols.get(id)) \|sym\| {
	1244	if (t.scopes.items[i].symbols.getPtr(id)) \|sym\| {
1243	1245	return sym;
1244	1246	}
1245	1247	}

src/compiler/main.zig+29-13

...	...	@@ -1,30 +1,40 @@
1	1	const std = @import("std");
	2	const builtin = @import("builtin");
	3
2	4	const Tokenizer = @import("Tokenizer.zig");
3	5	const Parse = @import("Parse.zig");
4	6	const Transform = @import("Transform.zig");
5	7	const Comp = @import("Comp.zig");
6	8	const unicode_props = @import("unicode_props.zig");
7		const T = @import("tracer");
8		pub const tracer_impl = @field(T, "none");
	9
	10	const tracer = @import("tracer");
	11
	12	pub const tracer_impl = @field(tracer, @import("build_options").tracing);
9	13
10	14	pub const Report = @import("Report.zig");
11	15
12	16	pub fn main() !void {
	17	try tracer.init();
	18	defer tracer.deinit();
	19	try tracer.init_thread(null);
	20	defer tracer.deinit_thread();
	21
13	22	try unicode_props.init();
14	23
15	24	std.debug.print("\r\x1b[K", .{});
16	25
17		try T.init();
18		defer T.deinit();
19
20		try T.init_thread(null);
21		defer T.deinit_thread();
	26	const use_gpa = builtin.mode == .Debug;
22	27
23		var gpa: std.heap.GeneralPurposeAllocator(.{
	28	var gpa: if (use_gpa) std.heap.GeneralPurposeAllocator(.{
24	29	.stack_trace_frames = 100,
25		}) = .{};
26		defer _ = gpa.deinit();
27		const alloc = gpa.allocator();
	30	}) else struct {} = .{};
	31	defer if (use_gpa) {
	32	_ = gpa.deinit();
	33	};
	34	const alloc = if (use_gpa)
	35	gpa.allocator()
	36	else
	37	std.heap.c_allocator;
28	38
29	39	var maybe_file_path: ?[]const u8 = null;
30	40	defer if (maybe_file_path) \|path\| alloc.free(path);
...	...	@@ -60,7 +70,7 @@ pub fn main() !void {
60	70	std.process.exit(1);
61	71	};
62	72
63		var comp: Comp = .{ .gpa = T.tracingAllocator(&alloc) };
	73	var comp: Comp = .{ .gpa = tracer.tracingAllocator(&alloc) };
64	74	try comp.init();
65	75	defer comp.deinit();
66	76	const main_file = try comp.getFile(file_path);
...	...	@@ -72,10 +82,14 @@ pub fn main() !void {
72	82	try fail_report.dumpToStdErr(&comp);
73	83	std.process.exit(1);
74	84	},
	85	error.ParseFail => {
	86	// Already printed.
	87	std.process.exit(1);
	88	},
75	89	else => return err,
76	90	};
77	91
78		{
	92	if (comp.options.dump_ir) {
79	93	const stderr = std.io.getStdErr().writer();
80	94	try teg.dumpToWriter(stderr);
81	95	}
...	...	@@ -88,6 +102,8 @@ pub fn main() !void {
88	102	try comp.link(writer);
89	103	try bufw.flush();
90	104
	105	std.process.cleanExit();
	106
91	107	// var error_pos: u32 = 0;
92	108	// var tokens = Tokenizer.tokenize(source, alloc.allocator(), &error_pos) catch \|err\| {
93	109	// var line: usize = 1;