authorgravatar for andrew@ziglang.orgAndrew Kelley <andrew@ziglang.org> 2020-04-23 16:41:20-04:00
committergravatar for andrew@ziglang.orgAndrew Kelley <andrew@ziglang.org> 2020-04-23 16:41:20-04:00
log24a01eed90b60a1e57172ae5a5305bc437bfeaba
treee7670e50bd61bcdeed198ec6f95b6ecdec33308d
parenta3dfe36ca1dac946f507c8b69241a93891bf7da5

basics of writing ELF and machine code generation

4 files changed, 357 insertions(+), 445 deletions(-)

src-self-hosted/codegen.zig+111-421
......@@ -1,447 +1,137 @@
11const std = @import("std");
2const Compilation = @import("compilation.zig").Compilation;
3const llvm = @import("llvm.zig");
4const c = @import("c.zig");
2const mem = std.mem;
3const assert = std.debug.assert;
54const ir = @import("ir.zig");
6const Value = @import("value.zig").Value;
75const Type = @import("type.zig").Type;
8const Scope = @import("scope.zig").Scope;
9const util = @import("util.zig");
10const event = std.event;
11const assert = std.debug.assert;
12const DW = std.dwarf;
13const maxInt = std.math.maxInt;
14
15pub async fn renderToLlvm(comp: *Compilation, fn_val: *Value.Fn, code: *ir.Code) Compilation.BuildError!void {
16 fn_val.base.ref();
17 defer fn_val.base.deref(comp);
18 defer code.destroy(comp.gpa());
19
20 var output_path = try comp.createRandomOutputPath(comp.target.oFileExt());
21 errdefer output_path.deinit();
22
23 const llvm_handle = try comp.zig_compiler.getAnyLlvmContext();
24 defer llvm_handle.release(comp.zig_compiler);
25
26 const context = llvm_handle.node.data;
27
28 const module = llvm.ModuleCreateWithNameInContext(comp.name.span(), context) orelse return error.OutOfMemory;
29 defer llvm.DisposeModule(module);
30
31 llvm.SetTarget(module, comp.llvm_triple.span());
32 llvm.SetDataLayout(module, comp.target_layout_str);
33
34 if (comp.target.getObjectFormat() == .coff) {
35 llvm.AddModuleCodeViewFlag(module);
36 } else {
37 llvm.AddModuleDebugInfoFlag(module);
38 }
39
40 const builder = llvm.CreateBuilderInContext(context) orelse return error.OutOfMemory;
41 defer llvm.DisposeBuilder(builder);
42
43 const dibuilder = llvm.CreateDIBuilder(module, true) orelse return error.OutOfMemory;
44 defer llvm.DisposeDIBuilder(dibuilder);
45
46 // Don't use ZIG_VERSION_STRING here. LLVM misparses it when it includes
47 // the git revision.
48 const producer = try std.fmt.allocPrintZ(&code.arena.allocator, "zig {}.{}.{}", .{
49 @as(u32, c.ZIG_VERSION_MAJOR),
50 @as(u32, c.ZIG_VERSION_MINOR),
51 @as(u32, c.ZIG_VERSION_PATCH),
52 });
53 const flags = "";
54 const runtime_version = 0;
55 const compile_unit_file = llvm.CreateFile(
56 dibuilder,
57 comp.name.span(),
58 comp.root_package.root_src_dir.span(),
59 ) orelse return error.OutOfMemory;
60 const is_optimized = comp.build_mode != .Debug;
61 const compile_unit = llvm.CreateCompileUnit(
62 dibuilder,
63 DW.LANG_C99,
64 compile_unit_file,
65 producer,
66 is_optimized,
67 flags,
68 runtime_version,
69 "",
70 0,
71 !comp.strip,
72 ) orelse return error.OutOfMemory;
73
74 var ofile = ObjectFile{
75 .comp = comp,
76 .module = module,
77 .builder = builder,
78 .dibuilder = dibuilder,
79 .context = context,
80 .lock = event.Lock.init(),
81 .arena = &code.arena.allocator,
82 };
83
84 try renderToLlvmModule(&ofile, fn_val, code);
85
86 // TODO module level assembly
87 //if (buf_len(&g->global_asm) != 0) {
88 // LLVMSetModuleInlineAsm(g->module, buf_ptr(&g->global_asm));
89 //}
90
91 llvm.DIBuilderFinalize(dibuilder);
92
93 if (comp.verbose_llvm_ir) {
94 std.debug.warn("raw module:\n", .{});
95 llvm.DumpModule(ofile.module);
96 }
6const Value = @import("value.zig").Value;
977
98 // verify the llvm module when safety is on
99 if (std.debug.runtime_safety) {
100 var error_ptr: ?[*:0]u8 = null;
101 _ = llvm.VerifyModule(ofile.module, llvm.AbortProcessAction, &error_ptr);
102 }
8pub const ErrorMsg = struct {
9 byte_offset: usize,
10 msg: []const u8,
11};
10312
104 const is_small = comp.build_mode == .ReleaseSmall;
105 const is_debug = comp.build_mode == .Debug;
13pub const Symbol = struct {
14 errors: []ErrorMsg,
10615
107 var err_msg: [*:0]u8 = undefined;
108 // TODO integrate this with evented I/O
109 if (llvm.TargetMachineEmitToFile(
110 comp.target_machine,
111 module,
112 output_path.span(),
113 llvm.EmitBinary,
114 &err_msg,
115 is_debug,
116 is_small,
117 )) {
118 if (std.debug.runtime_safety) {
119 std.debug.panic("unable to write object file {}: {s}\n", .{ output_path.span(), err_msg });
16 pub fn deinit(self: *Symbol, allocator: *mem.Allocator) void {
17 for (self.errors) |err| {
18 allocator.free(err.msg);
12019 }
121 return error.WritingObjectFileFailed;
122 }
123 //validate_inline_fns(g); TODO
124 fn_val.containing_object = output_path;
125 if (comp.verbose_llvm_ir) {
126 std.debug.warn("optimized module:\n", .{});
127 llvm.DumpModule(ofile.module);
128 }
129 if (comp.verbose_link) {
130 std.debug.warn("created {}\n", .{output_path.span()});
131 }
132}
133
134pub const ObjectFile = struct {
135 comp: *Compilation,
136 module: *llvm.Module,
137 builder: *llvm.Builder,
138 dibuilder: *llvm.DIBuilder,
139 context: *llvm.Context,
140 lock: event.Lock,
141 arena: *std.mem.Allocator,
142
143 fn gpa(self: *ObjectFile) *std.mem.Allocator {
144 return self.comp.gpa();
20 allocator.free(self.errors);
21 self.* = undefined;
14522 }
14623};
14724
148pub fn renderToLlvmModule(ofile: *ObjectFile, fn_val: *Value.Fn, code: *ir.Code) !void {
149 // TODO audit more of codegen.cpp:fn_llvm_value and port more logic
150 const llvm_fn_type = try fn_val.base.typ.getLlvmType(ofile.arena, ofile.context);
151 const llvm_fn = llvm.AddFunction(
152 ofile.module,
153 fn_val.symbol_name.span(),
154 llvm_fn_type,
155 ) orelse return error.OutOfMemory;
156
157 const want_fn_safety = fn_val.block_scope.?.safety.get(ofile.comp);
158 if (want_fn_safety and ofile.comp.haveLibC()) {
159 try addLLVMFnAttr(ofile, llvm_fn, "sspstrong");
160 try addLLVMFnAttrStr(ofile, llvm_fn, "stack-protector-buffer-size", "4");
25pub fn generateSymbol(typed_value: ir.TypedValue, module: ir.Module, code: *std.ArrayList(u8)) !Symbol {
26 switch (typed_value.ty.zigTypeTag()) {
27 .Fn => {
28 const index = typed_value.val.cast(Value.Payload.Function).?.index;
29 const module_fn = module.fns[index];
30
31 var function = Function{
32 .module = &module,
33 .mod_fn = &module_fn,
34 .code = code,
35 .inst_table = std.AutoHashMap(*ir.Inst, Function.MCValue).init(code.allocator),
36 .errors = std.ArrayList(ErrorMsg).init(code.allocator),
37 .constants = std.ArrayList(ir.TypedValue).init(code.allocator),
38 };
39 defer function.inst_table.deinit();
40 defer function.errors.deinit();
41
42 for (module_fn.body) |inst| {
43 const new_inst = function.genFuncInst(inst) catch |err| switch (err) {
44 error.CodegenFail => {
45 assert(function.errors.items.len != 0);
46 break;
47 },
48 else => |e| return e,
49 };
50 try function.inst_table.putNoClobber(inst, new_inst);
51 }
52 return Symbol{ .errors = function.errors.toOwnedSlice() };
53 },
54 else => @panic("TODO implement generateSymbol for non-function types"),
16155 }
56}
16257
163 // TODO
164 //if (fn_val.align_stack) |align_stack| {
165 // try addLLVMFnAttrInt(ofile, llvm_fn, "alignstack", align_stack);
166 //}
167
168 const fn_type = fn_val.base.typ.cast(Type.Fn).?;
169 const fn_type_normal = &fn_type.key.data.Normal;
170
171 try addLLVMFnAttr(ofile, llvm_fn, "nounwind");
172 //add_uwtable_attr(g, fn_table_entry->llvm_value);
173 try addLLVMFnAttr(ofile, llvm_fn, "nobuiltin");
174
175 //if (g->build_mode == BuildModeDebug && fn_table_entry->fn_inline != FnInlineAlways) {
176 // ZigLLVMAddFunctionAttr(fn_table_entry->llvm_value, "no-frame-pointer-elim", "true");
177 // ZigLLVMAddFunctionAttr(fn_table_entry->llvm_value, "no-frame-pointer-elim-non-leaf", nullptr);
178 //}
179
180 //if (fn_table_entry->section_name) {
181 // LLVMSetSection(fn_table_entry->llvm_value, buf_ptr(fn_table_entry->section_name));
182 //}
183 //if (fn_table_entry->align_bytes > 0) {
184 // LLVMSetAlignment(fn_table_entry->llvm_value, (unsigned)fn_table_entry->align_bytes);
185 //} else {
186 // // We'd like to set the best alignment for the function here, but on Darwin LLVM gives
187 // // "Cannot getTypeInfo() on a type that is unsized!" assertion failure when calling
188 // // any of the functions for getting alignment. Not specifying the alignment should
189 // // use the ABI alignment, which is fine.
190 //}
191
192 //if (!type_has_bits(return_type)) {
193 // // nothing to do
194 //} else if (type_is_codegen_pointer(return_type)) {
195 // addLLVMAttr(fn_table_entry->llvm_value, 0, "nonnull");
196 //} else if (handle_is_ptr(return_type) &&
197 // calling_convention_does_first_arg_return(fn_type->data.fn.fn_type_id.cc))
198 //{
199 // addLLVMArgAttr(fn_table_entry->llvm_value, 0, "sret");
200 // addLLVMArgAttr(fn_table_entry->llvm_value, 0, "nonnull");
201 //}
202
203 // TODO set parameter attributes
204
205 // TODO
206 //uint32_t err_ret_trace_arg_index = get_err_ret_trace_arg_index(g, fn_table_entry);
207 //if (err_ret_trace_arg_index != UINT32_MAX) {
208 // addLLVMArgAttr(fn_table_entry->llvm_value, (unsigned)err_ret_trace_arg_index, "nonnull");
209 //}
210
211 const cur_ret_ptr = if (fn_type_normal.return_type.handleIsPtr()) llvm.GetParam(llvm_fn, 0) else null;
212
213 // build all basic blocks
214 for (code.basic_block_list.span()) |bb| {
215 bb.llvm_block = llvm.AppendBasicBlockInContext(
216 ofile.context,
217 llvm_fn,
218 bb.name_hint,
219 ) orelse return error.OutOfMemory;
220 }
221 const entry_bb = code.basic_block_list.at(0);
222 llvm.PositionBuilderAtEnd(ofile.builder, entry_bb.llvm_block);
223
224 llvm.ClearCurrentDebugLocation(ofile.builder);
225
226 // TODO set up error return tracing
227 // TODO allocate temporary stack values
228
229 const var_list = fn_type.non_key.Normal.variable_list.span();
230 // create debug variable declarations for variables and allocate all local variables
231 for (var_list) |var_scope, i| {
232 const var_type = switch (var_scope.data) {
233 .Const => unreachable,
234 .Param => |param| param.typ,
235 };
236 // if (!type_has_bits(var->value->type)) {
237 // continue;
238 // }
239 // if (ir_get_var_is_comptime(var))
240 // continue;
241 // if (type_requires_comptime(var->value->type))
242 // continue;
243 // if (var->src_arg_index == SIZE_MAX) {
244 // var->value_ref = build_alloca(g, var->value->type, buf_ptr(&var->name), var->align_bytes);
245
246 // var->di_loc_var = ZigLLVMCreateAutoVariable(g->dbuilder, get_di_scope(g, var->parent_scope),
247 // buf_ptr(&var->name), import->di_file, (unsigned)(var->decl_node->line + 1),
248 // var->value->type->di_type, !g->strip_debug_symbols, 0);
249
250 // } else {
251 // it's a parameter
252 // assert(var->gen_arg_index != SIZE_MAX);
253 // TypeTableEntry *gen_type;
254 // FnGenParamInfo *gen_info = &fn_table_entry->type_entry->data.fn.gen_param_info[var->src_arg_index];
58const Function = struct {
59 module: *const ir.Module,
60 mod_fn: *const ir.Module.Fn,
61 code: *std.ArrayList(u8),
62 inst_table: std.AutoHashMap(*ir.Inst, MCValue),
63 /// Constants are embedded within functions (at the end, after `ret`)
64 /// so that they are independently updateable.
65 /// This is a list of constants that must be appended to the symbol after `ret`.
66 constants: std.ArrayList(ir.TypedValue),
67 errors: std.ArrayList(ErrorMsg),
68
69 const MCValue = union(enum) {
70 none,
71 unreach,
72 /// A pointer-sized integer that fits in a register.
73 immediate: u64,
74 /// Refers to the index into `constants` field of `Function`.
75 local_const_ptr: usize,
76 };
25577
256 if (var_type.handleIsPtr()) {
257 // if (gen_info->is_byval) {
258 // gen_type = var->value->type;
259 // } else {
260 // gen_type = gen_info->type;
261 // }
262 var_scope.data.Param.llvm_value = llvm.GetParam(llvm_fn, @intCast(c_uint, i));
263 } else {
264 // gen_type = var->value->type;
265 var_scope.data.Param.llvm_value = try renderAlloca(ofile, var_type, var_scope.name, .Abi);
78 fn genFuncInst(self: *Function, inst: *ir.Inst) !MCValue {
79 switch (inst.tag) {
80 .unreach => return self.genPanic(inst.src),
81 .constant => unreachable, // excluded from function bodies
82 .assembly => return self.genAsm(inst.cast(ir.Inst.Assembly).?),
83 .ptrtoint => return self.genPtrToInt(inst.cast(ir.Inst.PtrToInt).?),
26684 }
267 // if (var->decl_node) {
268 // var->di_loc_var = ZigLLVMCreateParameterVariable(g->dbuilder, get_di_scope(g, var->parent_scope),
269 // buf_ptr(&var->name), import->di_file,
270 // (unsigned)(var->decl_node->line + 1),
271 // gen_type->di_type, !g->strip_debug_symbols, 0, (unsigned)(var->gen_arg_index + 1));
272 // }
273
274 // }
27585 }
27686
277 // TODO finishing error return trace setup. we have to do this after all the allocas.
278
279 // create debug variable declarations for parameters
280 // rely on the first variables in the variable_list being parameters.
281 //size_t next_var_i = 0;
282 for (fn_type.key.data.Normal.params) |param, i| {
283 //FnGenParamInfo *info = &fn_table_entry->type_entry->data.fn.gen_param_info[param_i];
284 //if (info->gen_index == SIZE_MAX)
285 // continue;
286 const scope_var = var_list[i];
287 //assert(variable->src_arg_index != SIZE_MAX);
288 //next_var_i += 1;
289 //assert(variable);
290 //assert(variable->value_ref);
291
292 if (!param.typ.handleIsPtr()) {
293 //clear_debug_source_node(g);
294 const llvm_param = llvm.GetParam(llvm_fn, @intCast(c_uint, i));
295 _ = try renderStoreUntyped(
296 ofile,
297 llvm_param,
298 scope_var.data.Param.llvm_value,
299 .Abi,
300 .Non,
301 );
87 fn genPanic(self: *Function, src: usize) !MCValue {
88 // TODO change this to call the panic function
89 switch (self.module.target.cpu.arch) {
90 .i386, .x86_64 => {
91 try self.code.append(0xcc); // x86 int3
92 },
93 else => return self.fail(src, "TODO implement panic for {}", .{self.module.target.cpu.arch}),
30294 }
303
304 //if (variable->decl_node) {
305 // gen_var_debug_decl(g, variable);
306 //}
95 return .unreach;
30796 }
30897
309 for (code.basic_block_list.span()) |current_block| {
310 llvm.PositionBuilderAtEnd(ofile.builder, current_block.llvm_block);
311 for (current_block.instruction_list.span()) |instruction| {
312 if (instruction.ref_count == 0 and !instruction.hasSideEffects()) continue;
313
314 instruction.llvm_value = try instruction.render(ofile, fn_val);
315 }
316 current_block.llvm_exit_block = llvm.GetInsertBlock(ofile.builder);
98 fn genAsm(self: *Function, inst: *ir.Inst.Assembly) !MCValue {
99 return self.fail(inst.base.src, "TODO machine code gen assembly", .{});
317100 }
318}
319
320fn addLLVMAttr(
321 ofile: *ObjectFile,
322 val: *llvm.Value,
323 attr_index: llvm.AttributeIndex,
324 attr_name: []const u8,
325) !void {
326 const kind_id = llvm.GetEnumAttributeKindForName(attr_name.ptr, attr_name.len);
327 assert(kind_id != 0);
328 const llvm_attr = llvm.CreateEnumAttribute(ofile.context, kind_id, 0) orelse return error.OutOfMemory;
329 llvm.AddAttributeAtIndex(val, attr_index, llvm_attr);
330}
331
332fn addLLVMAttrStr(
333 ofile: *ObjectFile,
334 val: *llvm.Value,
335 attr_index: llvm.AttributeIndex,
336 attr_name: []const u8,
337 attr_val: []const u8,
338) !void {
339 const llvm_attr = llvm.CreateStringAttribute(
340 ofile.context,
341 attr_name.ptr,
342 @intCast(c_uint, attr_name.len),
343 attr_val.ptr,
344 @intCast(c_uint, attr_val.len),
345 ) orelse return error.OutOfMemory;
346 llvm.AddAttributeAtIndex(val, attr_index, llvm_attr);
347}
348
349fn addLLVMAttrInt(
350 val: *llvm.Value,
351 attr_index: llvm.AttributeIndex,
352 attr_name: []const u8,
353 attr_val: u64,
354) !void {
355 const kind_id = llvm.GetEnumAttributeKindForName(attr_name.ptr, attr_name.len);
356 assert(kind_id != 0);
357 const llvm_attr = llvm.CreateEnumAttribute(ofile.context, kind_id, attr_val) orelse return error.OutOfMemory;
358 llvm.AddAttributeAtIndex(val, attr_index, llvm_attr);
359}
360
361fn addLLVMFnAttr(ofile: *ObjectFile, fn_val: *llvm.Value, attr_name: []const u8) !void {
362 return addLLVMAttr(ofile, fn_val, maxInt(llvm.AttributeIndex), attr_name);
363}
364101
365fn addLLVMFnAttrStr(ofile: *ObjectFile, fn_val: *llvm.Value, attr_name: []const u8, attr_val: []const u8) !void {
366 return addLLVMAttrStr(ofile, fn_val, maxInt(llvm.AttributeIndex), attr_name, attr_val);
367}
368
369fn addLLVMFnAttrInt(ofile: *ObjectFile, fn_val: *llvm.Value, attr_name: []const u8, attr_val: u64) !void {
370 return addLLVMAttrInt(ofile, fn_val, maxInt(llvm.AttributeIndex), attr_name, attr_val);
371}
372
373fn renderLoadUntyped(
374 ofile: *ObjectFile,
375 ptr: *llvm.Value,
376 alignment: Type.Pointer.Align,
377 vol: Type.Pointer.Vol,
378 name: [*:0]const u8,
379) !*llvm.Value {
380 const result = llvm.BuildLoad(ofile.builder, ptr, name) orelse return error.OutOfMemory;
381 switch (vol) {
382 .Non => {},
383 .Volatile => llvm.SetVolatile(result, 1),
102 fn genPtrToInt(self: *Function, inst: *ir.Inst.PtrToInt) !MCValue {
103 // no-op
104 return self.resolveInst(inst.args.ptr);
384105 }
385 llvm.SetAlignment(result, resolveAlign(ofile, alignment, llvm.GetElementType(llvm.TypeOf(ptr))));
386 return result;
387}
388106
389fn renderLoad(ofile: *ObjectFile, ptr: *llvm.Value, ptr_type: *Type.Pointer, name: [*:0]const u8) !*llvm.Value {
390 return renderLoadUntyped(ofile, ptr, ptr_type.key.alignment, ptr_type.key.vol, name);
391}
392
393pub fn getHandleValue(ofile: *ObjectFile, ptr: *llvm.Value, ptr_type: *Type.Pointer) !?*llvm.Value {
394 const child_type = ptr_type.key.child_type;
395 if (!child_type.hasBits()) {
396 return null;
397 }
398 if (child_type.handleIsPtr()) {
399 return ptr;
107 fn resolveInst(self: *Function, inst: *ir.Inst) !MCValue {
108 if (inst.cast(ir.Inst.Constant)) |const_inst| {
109 switch (inst.ty.zigTypeTag()) {
110 .Int => {
111 const info = inst.ty.intInfo(self.module.target);
112 const ptr_bits = self.module.target.cpu.arch.ptrBitWidth();
113 if (info.bits > ptr_bits or info.signed) {
114 return self.fail(inst.src, "TODO const int bigger than ptr and signed int", .{});
115 }
116 return MCValue{ .immediate = const_inst.val.toUnsignedInt() };
117 },
118 else => return self.fail(inst.src, "TODO implement const of type '{}'", .{inst.ty}),
119 }
120 } else {
121 return self.inst_table.getValue(inst).?;
122 }
400123 }
401 return try renderLoad(ofile, ptr, ptr_type, "");
402}
403124
404pub fn renderStoreUntyped(
405 ofile: *ObjectFile,
406 value: *llvm.Value,
407 ptr: *llvm.Value,
408 alignment: Type.Pointer.Align,
409 vol: Type.Pointer.Vol,
410) !*llvm.Value {
411 const result = llvm.BuildStore(ofile.builder, value, ptr) orelse return error.OutOfMemory;
412 switch (vol) {
413 .Non => {},
414 .Volatile => llvm.SetVolatile(result, 1),
125 fn fail(self: *Function, src: usize, comptime format: []const u8, args: var) error{ CodegenFail, OutOfMemory } {
126 @setCold(true);
127 const msg = try std.fmt.allocPrint(self.errors.allocator, format, args);
128 {
129 errdefer self.errors.allocator.free(msg);
130 (try self.errors.addOne()).* = .{
131 .byte_offset = src,
132 .msg = msg,
133 };
134 }
135 return error.CodegenFail;
415136 }
416 llvm.SetAlignment(result, resolveAlign(ofile, alignment, llvm.TypeOf(value)));
417 return result;
418}
419
420pub fn renderStore(
421 ofile: *ObjectFile,
422 value: *llvm.Value,
423 ptr: *llvm.Value,
424 ptr_type: *Type.Pointer,
425) !*llvm.Value {
426 return renderStoreUntyped(ofile, value, ptr, ptr_type.key.alignment, ptr_type.key.vol);
427}
428
429pub fn renderAlloca(
430 ofile: *ObjectFile,
431 var_type: *Type,
432 name: []const u8,
433 alignment: Type.Pointer.Align,
434) !*llvm.Value {
435 const llvm_var_type = try var_type.getLlvmType(ofile.arena, ofile.context);
436 const name_with_null = try std.cstr.addNullByte(ofile.arena, name);
437 const result = llvm.BuildAlloca(ofile.builder, llvm_var_type, @ptrCast([*:0]const u8, name_with_null.ptr)) orelse return error.OutOfMemory;
438 llvm.SetAlignment(result, resolveAlign(ofile, alignment, llvm_var_type));
439 return result;
440}
441
442pub fn resolveAlign(ofile: *ObjectFile, alignment: Type.Pointer.Align, llvm_type: *llvm.Type) u32 {
443 return switch (alignment) {
444 .Abi => return llvm.ABIAlignmentOfType(ofile.comp.target_data_ref, llvm_type),
445 .Override => |a| a,
446 };
447}
137};
src-self-hosted/ir.zig+10-1
......@@ -724,7 +724,16 @@ pub fn main() anyerror!void {
724724 }
725725
726726 const link = @import("link.zig");
727 try link.updateExecutableFilePath(allocator, analyzed_module, std.fs.cwd(), "a.out");
727 var result = try link.updateExecutableFilePath(allocator, analyzed_module, std.fs.cwd(), "a.out");
728 defer result.deinit(allocator);
729 if (result.errors.len != 0) {
730 for (result.errors) |err_msg| {
731 const loc = findLineColumn(source, err_msg.byte_offset);
732 std.debug.warn("{}:{}:{}: error: {}\n", .{ src_path, loc.line + 1, loc.column + 1, err_msg.msg });
733 }
734 if (debug_error_trace) return error.ParseFailure;
735 std.process.exit(1);
736 }
728737}
729738
730739fn findLineColumn(source: []const u8, byte_offset: usize) struct { line: usize, column: usize } {
src-self-hosted/link.zig+186-23
......@@ -5,15 +5,38 @@ const Allocator = std.mem.Allocator;
55const ir = @import("ir.zig");
66const fs = std.fs;
77const elf = std.elf;
8const codegen = @import("codegen.zig");
89
910const executable_mode = 0o755;
1011const default_entry_addr = 0x8000000;
1112
13pub const ErrorMsg = struct {
14 byte_offset: usize,
15 msg: []const u8,
16};
17
18pub const Result = struct {
19 errors: []ErrorMsg,
20
21 pub fn deinit(self: *Result, allocator: *mem.Allocator) void {
22 for (self.errors) |err| {
23 allocator.free(err.msg);
24 }
25 allocator.free(self.errors);
26 self.* = undefined;
27 }
28};
29
1230/// Attempts incremental linking, if the file already exists.
1331/// If incremental linking fails, falls back to truncating the file and rewriting it.
1432/// A malicious file is detected as incremental link failure and does not cause Illegal Behavior.
1533/// This operation is not atomic.
16pub fn updateExecutableFilePath(allocator: *Allocator, module: ir.Module, dir: fs.Dir, sub_path: []const u8) !void {
34pub fn updateExecutableFilePath(
35 allocator: *Allocator,
36 module: ir.Module,
37 dir: fs.Dir,
38 sub_path: []const u8,
39) !Result {
1740 const file = try dir.createFile(sub_path, .{ .truncate = false, .read = true, .mode = executable_mode });
1841 defer file.close();
1942
......@@ -21,12 +44,18 @@ pub fn updateExecutableFilePath(allocator: *Allocator, module: ir.Module, dir: f
2144}
2245
2346/// Atomically overwrites the old file, if present.
24pub fn writeExecutableFilePath(allocator: *Allocator, module: ir.Module, dir: fs.Dir, sub_path: []const u8) !void {
47pub fn writeExecutableFilePath(
48 allocator: *Allocator,
49 module: ir.Module,
50 dir: fs.Dir,
51 sub_path: []const u8,
52) !Result {
2553 const af = try dir.atomicFile(sub_path, .{ .mode = executable_mode });
2654 defer af.deinit();
2755
28 try writeExecutableFile(allocator, module, af.file);
56 const result = try writeExecutableFile(allocator, module, af.file);
2957 try af.finish();
58 return result;
3059}
3160
3261/// Attempts incremental linking, if the file already exists.
......@@ -34,8 +63,8 @@ pub fn writeExecutableFilePath(allocator: *Allocator, module: ir.Module, dir: fs
3463/// Returns an error if `file` is not already open with +read +write +seek abilities.
3564/// A malicious file is detected as incremental link failure and does not cause Illegal Behavior.
3665/// This operation is not atomic.
37pub fn updateExecutableFile(allocator: *Allocator, module: ir.Module, file: fs.File) !void {
38 updateExecutableFileInner(allocator, module, file) catch |err| switch (err) {
66pub fn updateExecutableFile(allocator: *Allocator, module: ir.Module, file: fs.File) !Result {
67 return updateExecutableFileInner(allocator, module, file) catch |err| switch (err) {
3968 error.IncrFailed => {
4069 return writeExecutableFile(allocator, module, file);
4170 },
......@@ -66,20 +95,17 @@ const Update = struct {
6695 text_section_index: ?u16,
6796 symtab_section_index: ?u16,
6897
69 /// Key: index into strtab. Value: index into symbols.
70 symbol_table: std.AutoHashMap(usize, usize),
7198 /// The same order as in the file
7299 symbols: std.ArrayList(elf.Elf64_Sym),
73 /// Sorted by address, index into symbols
74 symbols_by_addr: std.ArrayList(usize),
100
101 errors: std.ArrayList(ErrorMsg),
75102
76103 fn deinit(self: *Update) void {
77104 self.sections.deinit();
78105 self.program_headers.deinit();
79106 self.shstrtab.deinit();
80 self.symbol_table.deinit();
81107 self.symbols.deinit();
82 self.symbols_by_addr.deinit();
108 self.errors.deinit();
83109 }
84110
85111 // `expand_num / expand_den` is the factor of padding when allocation
......@@ -162,6 +188,7 @@ const Update = struct {
162188 fn makeString(self: *Update, bytes: []const u8) !u32 {
163189 const result = self.shstrtab.items.len;
164190 try self.shstrtab.appendSlice(bytes);
191 try self.shstrtab.append(0);
165192 return @intCast(u32, result);
166193 }
167194
......@@ -187,6 +214,7 @@ const Update = struct {
187214 const file_size = 256 * 1024;
188215 const p_align = 0x1000;
189216 const off = self.findFreeSpace(file_size, p_align);
217 //std.debug.warn("found PT_LOAD free space 0x{x} to 0x{x}\n", .{ off, off + file_size });
190218 try self.program_headers.append(.{
191219 .p_type = elf.PT_LOAD,
192220 .p_offset = off,
......@@ -194,10 +222,10 @@ const Update = struct {
194222 .p_vaddr = default_entry_addr,
195223 .p_paddr = default_entry_addr,
196224 .p_memsz = 0,
197 .p_align = 0x1000,
225 .p_align = p_align,
198226 .p_flags = elf.PF_X | elf.PF_R,
199227 });
200 self.entry_addr = default_entry_addr;
228 self.entry_addr = null;
201229 phdr_load_re_dirty = true;
202230 phdr_table_dirty = true;
203231 }
......@@ -220,6 +248,7 @@ const Update = struct {
220248 if (self.shstrtab_index == null) {
221249 self.shstrtab_index = @intCast(u16, self.sections.items.len);
222250 const off = self.findFreeSpace(self.shstrtab.items.len, 1);
251 //std.debug.warn("found shstrtab free space 0x{x} to 0x{x}\n", .{ off, off + self.shstrtab.items.len });
223252 try self.sections.append(.{
224253 .sh_name = try self.makeString(".shstrtab"),
225254 .sh_type = elf.SHT_STRTAB,
......@@ -259,6 +288,7 @@ const Update = struct {
259288 const each_size: u64 = if (small_ptr) @sizeOf(elf.Elf32_Sym) else @sizeOf(elf.Elf64_Sym);
260289 const file_size = self.module.exports.len * each_size;
261290 const off = self.findFreeSpace(file_size, min_align);
291 //std.debug.warn("found symtab free space 0x{x} to 0x{x}\n", .{ off, off + file_size });
262292
263293 try self.sections.append(.{
264294 .sh_name = try self.makeString(".symtab"),
......@@ -307,6 +337,7 @@ const Update = struct {
307337 const needed_size = self.program_headers.items.len * phsize;
308338
309339 if (needed_size > allocated_size) {
340 self.phdr_table_offset = null; // free the space
310341 self.phdr_table_offset = self.findFreeSpace(needed_size, phalign);
311342 }
312343
......@@ -361,6 +392,7 @@ const Update = struct {
361392 const needed_size = self.sections.items.len * phsize;
362393
363394 if (needed_size > allocated_size) {
395 self.shdr_table_offset = null; // free the space
364396 self.shdr_table_offset = self.findFreeSpace(needed_size, phalign);
365397 }
366398
......@@ -414,11 +446,30 @@ const Update = struct {
414446 },
415447 }
416448 }
417 if (shstrtab_dirty) {
418 try self.file.pwriteAll(self.shstrtab.items, self.sections.items[self.shstrtab_index.?].sh_offset);
419 }
420449 try self.writeCodeAndSymbols();
421 try self.writeElfHeader();
450
451 const shstrtab_sect = &self.sections.items[self.shstrtab_index.?];
452 if (shstrtab_dirty or self.shstrtab.items.len != shstrtab_sect.sh_size) {
453 const allocated_size = self.allocatedSize(shstrtab_sect.sh_offset);
454 const needed_size = self.shstrtab.items.len;
455
456 if (needed_size > allocated_size) {
457 shstrtab_sect.sh_size = 0; // free the space
458 shstrtab_sect.sh_offset = self.findFreeSpace(needed_size, 1);
459 shstrtab_sect.sh_size = needed_size;
460 }
461 try self.file.pwriteAll(self.shstrtab.items, shstrtab_sect.sh_offset);
462 }
463 if (self.entry_addr == null) {
464 const msg = try std.fmt.allocPrint(self.errors.allocator, "no entry point found", .{});
465 errdefer self.errors.allocator.free(msg);
466 try self.errors.append(.{
467 .byte_offset = 0,
468 .msg = msg,
469 });
470 } else {
471 try self.writeElfHeader();
472 }
422473 // TODO find end pos and truncate
423474 }
424475
......@@ -540,13 +591,122 @@ const Update = struct {
540591 }
541592
542593 fn writeCodeAndSymbols(self: *Update) !void {
543 @panic("TODO writeCodeAndSymbols");
594 // index 0 is always a null symbol
595 try self.symbols.resize(1);
596 self.symbols.items[0] = .{
597 .st_name = 0,
598 .st_info = 0,
599 .st_other = 0,
600 .st_shndx = 0,
601 .st_value = 0,
602 .st_size = 0,
603 };
604
605 const phdr = &self.program_headers.items[self.phdr_load_re_index.?];
606 var vaddr: u64 = phdr.p_vaddr;
607
608 var code = std.ArrayList(u8).init(self.sections.allocator);
609 defer code.deinit();
610
611 for (self.module.exports) |exp| {
612 code.shrink(0);
613 var symbol = try codegen.generateSymbol(exp.typed_value, self.module.*, &code);
614 defer symbol.deinit(code.allocator);
615 if (symbol.errors.len != 0) {
616 for (symbol.errors) |err| {
617 const msg = try mem.dupe(self.errors.allocator, u8, err.msg);
618 errdefer self.errors.allocator.free(msg);
619 try self.errors.append(.{
620 .byte_offset = err.byte_offset,
621 .msg = msg,
622 });
623 }
624 continue;
625 }
626
627 if (mem.eql(u8, exp.name, "_start")) {
628 self.entry_addr = vaddr;
629 }
630 (try self.symbols.addOne()).* = .{
631 .st_name = try self.makeString(exp.name),
632 .st_info = (elf.STB_LOCAL << 4) | elf.STT_FUNC,
633 .st_other = 0,
634 .st_shndx = self.text_section_index.?,
635 .st_value = vaddr,
636 .st_size = code.items.len,
637 };
638 vaddr += code.items.len;
639 }
640
641 return self.writeSymbols();
642 }
643
644 fn writeSymbols(self: *Update) !void {
645 const ptr_width: enum { p32, p64 } = switch (self.module.target.cpu.arch.ptrBitWidth()) {
646 32 => .p32,
647 64 => .p64,
648 else => return error.UnsupportedArchitecture,
649 };
650 const small_ptr = ptr_width == .p32;
651 const syms_sect = &self.sections.items[self.symtab_section_index.?];
652 const sym_align: u16 = if (small_ptr) @alignOf(elf.Elf32_Sym) else @alignOf(elf.Elf64_Sym);
653 const sym_size: u64 = if (small_ptr) @sizeOf(elf.Elf32_Sym) else @sizeOf(elf.Elf64_Sym);
654
655 const allocated_size = self.allocatedSize(syms_sect.sh_offset);
656 const needed_size = self.symbols.items.len * sym_size;
657 if (needed_size > allocated_size) {
658 syms_sect.sh_size = 0; // free the space
659 syms_sect.sh_offset = self.findFreeSpace(needed_size, sym_align);
660 syms_sect.sh_size = needed_size;
661 }
662 const allocator = self.symbols.allocator;
663 const foreign_endian = self.module.target.cpu.arch.endian() != std.Target.current.cpu.arch.endian();
664 switch (ptr_width) {
665 .p32 => {
666 const buf = try allocator.alloc(elf.Elf32_Sym, self.symbols.items.len);
667 defer allocator.free(buf);
668
669 for (buf) |*sym, i| {
670 sym.* = .{
671 .st_name = self.symbols.items[i].st_name,
672 .st_value = @intCast(u32, self.symbols.items[i].st_value),
673 .st_size = @intCast(u32, self.symbols.items[i].st_size),
674 .st_info = self.symbols.items[i].st_info,
675 .st_other = self.symbols.items[i].st_other,
676 .st_shndx = self.symbols.items[i].st_shndx,
677 };
678 if (foreign_endian) {
679 bswapAllFields(elf.Elf32_Sym, sym);
680 }
681 }
682 try self.file.pwriteAll(mem.sliceAsBytes(buf), syms_sect.sh_offset);
683 },
684 .p64 => {
685 const buf = try allocator.alloc(elf.Elf64_Sym, self.symbols.items.len);
686 defer allocator.free(buf);
687
688 for (buf) |*sym, i| {
689 sym.* = .{
690 .st_name = self.symbols.items[i].st_name,
691 .st_value = self.symbols.items[i].st_value,
692 .st_size = self.symbols.items[i].st_size,
693 .st_info = self.symbols.items[i].st_info,
694 .st_other = self.symbols.items[i].st_other,
695 .st_shndx = self.symbols.items[i].st_shndx,
696 };
697 if (foreign_endian) {
698 bswapAllFields(elf.Elf64_Sym, sym);
699 }
700 }
701 try self.file.pwriteAll(mem.sliceAsBytes(buf), syms_sect.sh_offset);
702 },
703 }
544704 }
545705};
546706
547707/// Truncates the existing file contents and overwrites the contents.
548708/// Returns an error if `file` is not already open with +read +write +seek abilities.
549pub fn writeExecutableFile(allocator: *Allocator, module: ir.Module, file: fs.File) !void {
709pub fn writeExecutableFile(allocator: *Allocator, module: ir.Module, file: fs.File) !Result {
550710 var update = Update{
551711 .file = file,
552712 .module = &module,
......@@ -561,17 +721,20 @@ pub fn writeExecutableFile(allocator: *Allocator, module: ir.Module, file: fs.Fi
561721 .text_section_index = null,
562722 .symtab_section_index = null,
563723
564 .symbol_table = std.AutoHashMap(usize, usize).init(allocator),
565724 .symbols = std.ArrayList(elf.Elf64_Sym).init(allocator),
566 .symbols_by_addr = std.ArrayList(usize).init(allocator),
725
726 .errors = std.ArrayList(ErrorMsg).init(allocator),
567727 };
568728 defer update.deinit();
569729
570 return update.perform();
730 try update.perform();
731 return Result{
732 .errors = update.errors.toOwnedSlice(),
733 };
571734}
572735
573736/// Returns error.IncrFailed if incremental update could not be performed.
574fn updateExecutableFileInner(allocator: *Allocator, module: ir.Module, file: fs.File) !void {
737fn updateExecutableFileInner(allocator: *Allocator, module: ir.Module, file: fs.File) !Result {
575738 //var ehdr_buf: [@sizeOf(elf.Elf64_Ehdr)]u8 = undefined;
576739
577740 // TODO implement incremental linking
src-self-hosted/value.zig+50
......@@ -264,6 +264,56 @@ pub const Value = extern union {
264264 }
265265 }
266266
267 /// Asserts the value is an integer and it fits in a u64
268 pub fn toUnsignedInt(self: Value) u64 {
269 switch (self.tag()) {
270 .ty,
271 .u8_type,
272 .i8_type,
273 .isize_type,
274 .usize_type,
275 .c_short_type,
276 .c_ushort_type,
277 .c_int_type,
278 .c_uint_type,
279 .c_long_type,
280 .c_ulong_type,
281 .c_longlong_type,
282 .c_ulonglong_type,
283 .c_longdouble_type,
284 .f16_type,
285 .f32_type,
286 .f64_type,
287 .f128_type,
288 .c_void_type,
289 .bool_type,
290 .void_type,
291 .type_type,
292 .anyerror_type,
293 .comptime_int_type,
294 .comptime_float_type,
295 .noreturn_type,
296 .fn_naked_noreturn_no_args_type,
297 .single_const_pointer_to_comptime_int_type,
298 .const_slice_u8_type,
299 .void_value,
300 .noreturn_value,
301 .bool_true,
302 .bool_false,
303 .function,
304 .ref,
305 .ref_val,
306 .bytes,
307 => unreachable,
308
309 .zero => return 0,
310
311 .int_u64 => return self.cast(Payload.Int_u64).?.int,
312 .int_i64 => return @intCast(u64, self.cast(Payload.Int_u64).?.int),
313 .int_big => return self.cast(Payload.IntBig).?.big_int.to(u64) catch unreachable,
314 }
315 }
316
267317 /// Asserts the value is an integer, and the destination type is ComptimeInt or Int.
268318 pub fn intFitsInType(self: Value, ty: Type, target: Target) bool {
269319 switch (self.tag()) {