Commit c99ef23862

Luuk de Gram <luuk@degram.dev>
2024-02-07 06:57:32
wasm: consolidate flushModule and linkWithZld
We now use a single function to use the in-house WebAssembly linker rather than wasm-ld. For both incremental compilation and traditional linking we use the same codepath.
1 parent 5aec88f
Changed files (2)
src
src/link/Wasm/ZigObject.zig
@@ -670,17 +670,15 @@ pub fn addOrUpdateImport(
 
     if (type_index) |ty_index| {
         const gop = try zig_object.imports.getOrPut(gpa, symbol_index);
-        const module_name = if (lib_name) |l_name| blk: {
-            break :blk l_name;
-        } else wasm_file.host_name;
+        const module_name = if (lib_name) |l_name| l_name else wasm_file.host_name;
         if (!gop.found_existing) {
-            gop.value_ptr.* = .{
-                .module_name = try zig_object.string_table.insert(gpa, module_name),
-                .name = try zig_object.string_table.insert(gpa, name),
-                .kind = .{ .function = ty_index },
-            };
             zig_object.imported_functions_count += 1;
         }
+        gop.value_ptr.* = .{
+            .module_name = try zig_object.string_table.insert(gpa, module_name),
+            .name = try zig_object.string_table.insert(gpa, name),
+            .kind = .{ .function = ty_index },
+        };
         sym.tag = .function;
     } else {
         sym.tag = .data;
src/link/Wasm.zig
@@ -177,10 +177,6 @@ undefs: std.AutoArrayHashMapUnmanaged(u32, SymbolLoc) = .{},
 /// data of a symbol, such as its size, or its offset to perform a relocation.
 /// Undefined (and synthetic) symbols do not have an Atom and therefore cannot be mapped.
 symbol_atom: std.AutoHashMapUnmanaged(SymbolLoc, Atom.Index) = .{},
-/// Maps a symbol's location to its export name, which may differ from the decl's name
-/// which does the exporting.
-/// Note: The value represents the offset into the string table, rather than the actual string.
-export_names: std.AutoHashMapUnmanaged(SymbolLoc, u32) = .{},
 
 /// List of atom indexes of functions that are generated by the backend,
 /// rather than by the linker.
@@ -1398,7 +1394,6 @@ pub fn deinit(wasm: *Wasm) void {
     wasm.undefs.deinit(gpa);
     wasm.discarded.deinit(gpa);
     wasm.symbol_atom.deinit(gpa);
-    wasm.export_names.deinit(gpa);
     wasm.atoms.deinit(gpa);
     wasm.managed_atoms.deinit(gpa);
     wasm.segments.deinit(gpa);
@@ -2133,10 +2128,10 @@ fn setupExports(wasm: *Wasm) !void {
         if (!symbol.isExported(comp.config.rdynamic)) continue;
 
         const sym_name = sym_loc.getName(wasm);
-        const export_name = if (wasm.export_names.get(sym_loc)) |name| name else blk: {
-            if (sym_loc.file == .null) break :blk symbol.name;
-            break :blk try wasm.string_table.put(gpa, sym_name);
-        };
+        const export_name = if (sym_loc.file == .null)
+            symbol.name
+        else
+            try wasm.string_table.put(gpa, sym_name);
         const exp: types.Export = if (symbol.tag == .data) exp: {
             const global_index = @as(u32, @intCast(wasm.imported_globals_count + wasm.wasm_globals.items.len));
             try wasm.wasm_globals.append(gpa, .{
@@ -2437,168 +2432,45 @@ fn appendDummySegment(wasm: *Wasm) !void {
     });
 }
 
-fn resetState(wasm: *Wasm) void {
-    const gpa = wasm.base.comp.gpa;
-
-    for (wasm.segment_info.values()) |segment_info| {
-        gpa.free(segment_info.name);
-    }
-
-    // TODO: Revisit
-    // var atom_it = wasm.decls.valueIterator();
-    // while (atom_it.next()) |atom_index| {
-    //     const atom = wasm.getAtomPtr(atom_index.*);
-    //     atom.next = null;
-    //     atom.prev = null;
-
-    //     for (atom.locals.items) |local_atom_index| {
-    //         const local_atom = wasm.getAtomPtr(local_atom_index);
-    //         local_atom.next = null;
-    //         local_atom.prev = null;
-    //     }
-    // }
-
-    wasm.functions.clearRetainingCapacity();
-    wasm.exports.clearRetainingCapacity();
-    wasm.segments.clearRetainingCapacity();
-    wasm.segment_info.clearRetainingCapacity();
-    wasm.data_segments.clearRetainingCapacity();
-    wasm.atoms.clearRetainingCapacity();
-    wasm.symbol_atom.clearRetainingCapacity();
-    wasm.code_section_index = null;
-    wasm.debug_info_index = null;
-    wasm.debug_line_index = null;
-    wasm.debug_loc_index = null;
-    wasm.debug_str_index = null;
-    wasm.debug_ranges_index = null;
-    wasm.debug_abbrev_index = null;
-    wasm.debug_pubnames_index = null;
-    wasm.debug_pubtypes_index = null;
-}
-
 pub fn flush(wasm: *Wasm, arena: Allocator, prog_node: *std.Progress.Node) link.File.FlushError!void {
     const comp = wasm.base.comp;
     const use_lld = build_options.have_llvm and comp.config.use_lld;
-    const use_llvm = comp.config.use_llvm;
 
     if (use_lld) {
         return wasm.linkWithLLD(arena, prog_node);
-    } else if (use_llvm) {
-        return wasm.linkWithZld(arena, prog_node);
-    } else {
-        return wasm.flushModule(arena, prog_node);
     }
+    return wasm.flushModule(arena, prog_node);
 }
 
 /// Uses the in-house linker to link one or multiple object -and archive files into a WebAssembly binary.
-fn linkWithZld(wasm: *Wasm, arena: Allocator, prog_node: *std.Progress.Node) link.File.FlushError!void {
+pub fn flushModule(wasm: *Wasm, arena: Allocator, prog_node: *std.Progress.Node) link.File.FlushError!void {
     const tracy = trace(@src());
     defer tracy.end();
 
     const comp = wasm.base.comp;
-    const shared_memory = comp.config.shared_memory;
-    const import_memory = comp.config.import_memory;
+    if (wasm.llvm_object) |llvm_object| {
+        try wasm.base.emitLlvmObject(arena, llvm_object, prog_node);
+        const use_lld = build_options.have_llvm and comp.config.use_lld;
+        if (use_lld) return;
+    }
+
+    var sub_prog_node = prog_node.start("Wasm Flush", 0);
+    sub_prog_node.activate();
+    defer sub_prog_node.end();
 
     const directory = wasm.base.emit.directory; // Just an alias to make it shorter to type.
     const full_out_path = try directory.join(arena, &[_][]const u8{wasm.base.emit.sub_path});
-    const opt_zcu = comp.module;
-    const use_llvm = comp.config.use_llvm;
-
-    // If there is no Zig code to compile, then we should skip flushing the output file because it
-    // will not be part of the linker line anyway.
-    const module_obj_path: ?[]const u8 = if (opt_zcu != null) blk: {
-        assert(use_llvm); // `linkWithZld` should never be called when the Wasm backend is used
-        try wasm.flushModule(arena, prog_node);
-
+    const module_obj_path: ?[]const u8 = if (wasm.base.zcu_object_sub_path) |path| blk: {
         if (fs.path.dirname(full_out_path)) |dirname| {
-            break :blk try fs.path.join(arena, &.{ dirname, wasm.base.zcu_object_sub_path.? });
+            break :blk try fs.path.join(arena, &.{ dirname, path });
         } else {
-            break :blk wasm.base.zcu_object_sub_path.?;
+            break :blk path;
         }
     } else null;
 
-    var sub_prog_node = prog_node.start("Wasm Flush", 0);
-    sub_prog_node.activate();
-    defer sub_prog_node.end();
-
-    const compiler_rt_path: ?[]const u8 = blk: {
-        if (comp.compiler_rt_obj) |obj| break :blk obj.full_object_path;
-        if (comp.compiler_rt_lib) |lib| break :blk lib.full_object_path;
-        break :blk null;
-    };
-
-    const id_symlink_basename = "zld.id";
-
-    var man: Cache.Manifest = undefined;
-    defer if (!wasm.base.disable_lld_caching) man.deinit();
-    var digest: [Cache.hex_digest_len]u8 = undefined;
-
-    const objects = comp.objects;
-
-    // NOTE: The following section must be maintained to be equal
-    // as the section defined in `linkWithLLD`
-    if (!wasm.base.disable_lld_caching) {
-        man = comp.cache_parent.obtain();
-
-        // We are about to obtain this lock, so here we give other processes a chance first.
-        wasm.base.releaseLock();
-
-        comptime assert(Compilation.link_hash_implementation_version == 12);
-
-        for (objects) |obj| {
-            _ = try man.addFile(obj.path, null);
-            man.hash.add(obj.must_link);
-        }
-        for (comp.c_object_table.keys()) |key| {
-            _ = try man.addFile(key.status.success.object_path, null);
-        }
-        try man.addOptionalFile(module_obj_path);
-        try man.addOptionalFile(compiler_rt_path);
-        man.hash.addOptionalBytes(wasm.entry_name);
-        man.hash.add(wasm.base.stack_size);
-        man.hash.add(wasm.base.build_id);
-        man.hash.add(import_memory);
-        man.hash.add(shared_memory);
-        man.hash.add(wasm.import_table);
-        man.hash.add(wasm.export_table);
-        man.hash.addOptional(wasm.initial_memory);
-        man.hash.addOptional(wasm.max_memory);
-        man.hash.addOptional(wasm.global_base);
-        man.hash.addListOfBytes(wasm.export_symbol_names);
-        // strip does not need to go into the linker hash because it is part of the hash namespace
-
-        // We don't actually care whether it's a cache hit or miss; we just need the digest and the lock.
-        _ = try man.hit();
-        digest = man.final();
-
-        var prev_digest_buf: [digest.len]u8 = undefined;
-        const prev_digest: []u8 = Cache.readSmallFile(
-            directory.handle,
-            id_symlink_basename,
-            &prev_digest_buf,
-        ) catch |err| blk: {
-            log.debug("WASM LLD new_digest={s} error: {s}", .{ std.fmt.fmtSliceHexLower(&digest), @errorName(err) });
-            // Handle this as a cache miss.
-            break :blk prev_digest_buf[0..0];
-        };
-        if (mem.eql(u8, prev_digest, &digest)) {
-            log.debug("WASM LLD digest={s} match - skipping invocation", .{std.fmt.fmtSliceHexLower(&digest)});
-            // Hot diggity dog! The output binary is already there.
-            wasm.base.lock = man.toOwnedLock();
-            return;
-        }
-        log.debug("WASM LLD prev_digest={s} new_digest={s}", .{ std.fmt.fmtSliceHexLower(prev_digest), std.fmt.fmtSliceHexLower(&digest) });
-
-        // We are about to change the output file to be different, so we invalidate the build hash now.
-        directory.handle.deleteFile(id_symlink_basename) catch |err| switch (err) {
-            error.FileNotFound => {},
-            else => |e| return e,
-        };
-    }
-
     // Positional arguments to the linker such as object files and static archives.
     var positionals = std.ArrayList([]const u8).init(arena);
-    try positionals.ensureUnusedCapacity(objects.len);
+    try positionals.ensureUnusedCapacity(comp.objects.len);
 
     const target = comp.root_mod.resolved_target.result;
     const output_mode = comp.config.output_mode;
@@ -2607,6 +2479,10 @@ fn linkWithZld(wasm: *Wasm, arena: Allocator, prog_node: *std.Progress.Node) lin
     const link_libcpp = comp.config.link_libcpp;
     const wasi_exec_model = comp.config.wasi_exec_model;
 
+    if (wasm.zigObjectPtr()) |zig_object| {
+        try zig_object.flushModule(wasm);
+    }
+
     // When the target os is WASI, we allow linking with WASI-LIBC
     if (target.os.tag == .wasi) {
         const is_exe_or_dyn_lib = output_mode == .Exe or
@@ -2638,7 +2514,7 @@ fn linkWithZld(wasm: *Wasm, arena: Allocator, prog_node: *std.Progress.Node) lin
         try positionals.append(path);
     }
 
-    for (objects) |object| {
+    for (comp.objects) |object| {
         try positionals.append(object.path);
     }
 
@@ -2651,93 +2527,6 @@ fn linkWithZld(wasm: *Wasm, arena: Allocator, prog_node: *std.Progress.Node) lin
 
     try wasm.parseInputFiles(positionals.items);
 
-    for (wasm.objects.items) |object_index| {
-        try wasm.resolveSymbolsInObject(object_index);
-    }
-
-    var emit_features_count: u32 = 0;
-    var enabled_features: [@typeInfo(types.Feature.Tag).Enum.fields.len]bool = undefined;
-    try wasm.validateFeatures(&enabled_features, &emit_features_count);
-    try wasm.resolveSymbolsInArchives();
-    try wasm.resolveLazySymbols();
-    try wasm.checkUndefinedSymbols();
-
-    try wasm.setupInitFunctions();
-    try wasm.setupStart();
-
-    try wasm.markReferences();
-    try wasm.setupImports();
-    try wasm.mergeSections();
-    try wasm.mergeTypes();
-    try wasm.allocateAtoms();
-    try wasm.setupMemory();
-    wasm.allocateVirtualAddresses();
-    wasm.mapFunctionTable();
-    try wasm.initializeCallCtorsFunction();
-    try wasm.setupInitMemoryFunction();
-    try wasm.setupTLSRelocationsFunction();
-    try wasm.initializeTLSFunction();
-    try wasm.setupStartSection();
-    try wasm.setupExports();
-    try wasm.writeToFile(enabled_features, emit_features_count, arena);
-
-    if (!wasm.base.disable_lld_caching) {
-        // Update the file with the digest. If it fails we can continue; it only
-        // means that the next invocation will have an unnecessary cache miss.
-        Cache.writeSmallFile(directory.handle, id_symlink_basename, &digest) catch |err| {
-            log.warn("failed to save linking hash digest symlink: {s}", .{@errorName(err)});
-        };
-        // Again failure here only means an unnecessary cache miss.
-        man.writeManifest() catch |err| {
-            log.warn("failed to write cache manifest when linking: {s}", .{@errorName(err)});
-        };
-        // We hang on to this lock so that the output file path can be used without
-        // other processes clobbering it.
-        wasm.base.lock = man.toOwnedLock();
-    }
-}
-
-pub fn flushModule(wasm: *Wasm, arena: Allocator, prog_node: *std.Progress.Node) link.File.FlushError!void {
-    const tracy = trace(@src());
-    defer tracy.end();
-
-    const comp = wasm.base.comp;
-
-    if (wasm.llvm_object) |llvm_object| {
-        try wasm.base.emitLlvmObject(arena, llvm_object, prog_node);
-        return;
-    }
-
-    var sub_prog_node = prog_node.start("Wasm Flush", 0);
-    sub_prog_node.activate();
-    defer sub_prog_node.end();
-
-    if (wasm.zigObjectPtr()) |zig_object| {
-        try zig_object.flushModule(wasm);
-    }
-
-    // ensure the error names table is populated when an error name is referenced
-    // try wasm.populateErrorNameTable();
-
-    const objects = comp.objects;
-
-    // Positional arguments to the linker such as object files and static archives.
-    var positionals = std.ArrayList([]const u8).init(arena);
-    try positionals.ensureUnusedCapacity(objects.len);
-
-    for (objects) |object| {
-        positionals.appendAssumeCapacity(object.path);
-    }
-
-    for (comp.c_object_table.keys()) |c_object| {
-        try positionals.append(c_object.status.success.object_path);
-    }
-
-    if (comp.compiler_rt_lib) |lib| try positionals.append(lib.full_object_path);
-    if (comp.compiler_rt_obj) |obj| try positionals.append(obj.full_object_path);
-
-    try wasm.parseInputFiles(positionals.items);
-
     if (wasm.zig_object_index != .null) {
         try wasm.resolveSymbolsInObject(wasm.zig_object_index);
     }
@@ -2752,73 +2541,11 @@ pub fn flushModule(wasm: *Wasm, arena: Allocator, prog_node: *std.Progress.Node)
     try wasm.resolveLazySymbols();
     try wasm.checkUndefinedSymbols();
 
-    // When we finish/error we reset the state of the linker
-    // So we can rebuild the binary file on each incremental update
-    defer wasm.resetState();
     try wasm.setupInitFunctions();
     try wasm.setupStart();
+
     try wasm.markReferences();
-    // try wasm.setupErrorsLen();
     try wasm.setupImports();
-    // if (comp.module) |mod| {
-    //     var decl_it = wasm.decls.iterator();
-    //     while (decl_it.next()) |entry| {
-    //         const decl = mod.declPtr(entry.key_ptr.*);
-    //         if (decl.isExtern(mod)) continue;
-    //         const atom_index = entry.value_ptr.*;
-    //         const atom = wasm.getAtomPtr(atom_index);
-    //         if (decl.ty.zigTypeTag(mod) == .Fn) {
-    //             try wasm.parseAtom(atom_index, .function);
-    //         } else if (decl.getOwnedVariable(mod)) |variable| {
-    //             if (variable.is_const) {
-    //                 try wasm.parseAtom(atom_index, .{ .data = .read_only });
-    //             } else if (Value.fromInterned(variable.init).isUndefDeep(mod)) {
-    //                 // for safe build modes, we store the atom in the data segment,
-    //                 // whereas for unsafe build modes we store it in bss.
-    //                 const decl_namespace = mod.namespacePtr(decl.src_namespace);
-    //                 const optimize_mode = decl_namespace.file_scope.mod.optimize_mode;
-    //                 const is_initialized = switch (optimize_mode) {
-    //                     .Debug, .ReleaseSafe => true,
-    //                     .ReleaseFast, .ReleaseSmall => false,
-    //                 };
-    //                 try wasm.parseAtom(atom_index, .{ .data = if (is_initialized) .initialized else .uninitialized });
-    //             } else {
-    //                 // when the decl is all zeroes, we store the atom in the bss segment,
-    //                 // in all other cases it will be in the data segment.
-    //                 const is_zeroes = for (atom.code.items) |byte| {
-    //                     if (byte != 0) break false;
-    //                 } else true;
-    //                 try wasm.parseAtom(atom_index, .{ .data = if (is_zeroes) .uninitialized else .initialized });
-    //             }
-    //         } else {
-    //             try wasm.parseAtom(atom_index, .{ .data = .read_only });
-    //         }
-
-    //         // also parse atoms for a decl's locals
-    //         for (atom.locals.items) |local_atom_index| {
-    //             try wasm.parseAtom(local_atom_index, .{ .data = .read_only });
-    //         }
-    //     }
-    //     // parse anonymous declarations
-    //     for (wasm.anon_decls.keys(), wasm.anon_decls.values()) |decl_val, atom_index| {
-    //         const ty = Type.fromInterned(mod.intern_pool.typeOf(decl_val));
-    //         if (ty.zigTypeTag(mod) == .Fn) {
-    //             try wasm.parseAtom(atom_index, .function);
-    //         } else {
-    //             try wasm.parseAtom(atom_index, .{ .data = .read_only });
-    //         }
-    //     }
-
-    //     // also parse any backend-generated functions
-    //     for (wasm.synthetic_functions.items) |atom_index| {
-    //         try wasm.parseAtom(atom_index, .function);
-    //     }
-
-    //     if (wasm.dwarf) |*dwarf| {
-    //         try dwarf.flushModule(comp.module.?);
-    //     }
-    // }
-
     try wasm.mergeSections();
     try wasm.mergeTypes();
     try wasm.allocateAtoms();
@@ -4032,7 +3759,7 @@ fn emitSymbolTable(wasm: *Wasm, binary_bytes: *std.ArrayList(u8), symbol_table:
         try leb.writeULEB128(writer, @intFromEnum(symbol.tag));
         try leb.writeULEB128(writer, symbol.flags);
 
-        const sym_name = if (wasm.export_names.get(sym_loc)) |exp_name| wasm.string_table.get(exp_name) else sym_loc.getName(wasm);
+        const sym_name = sym_loc.getName(wasm);
         switch (symbol.tag) {
             .data => {
                 try leb.writeULEB128(writer, @as(u32, @intCast(sym_name.len)));