Merge pull request #11875 from motiejus/pagezero-size

[MachO] add -pagezero_size

src/Compilation.zig

@@ -903,6 +903,8 @@ pub const InitOptions = struct {
     install_name: ?[]const u8 = null,
     /// (Darwin) Path to entitlements file
     entitlements: ?[]const u8 = null,
+    /// (Darwin) size of the __PAGEZERO segment
+    pagezero_size: ?u64 = null,
 };
 
 fn addPackageTableToCacheHash(
@@ -1742,6 +1744,7 @@ pub fn create(gpa: Allocator, options: InitOptions) !*Compilation {
             .native_darwin_sdk = options.native_darwin_sdk,
             .install_name = options.install_name,
             .entitlements = options.entitlements,
+            .pagezero_size = options.pagezero_size,
         });
         errdefer bin_file.destroy();
         comp.* = .{
@@ -2359,7 +2362,7 @@ fn prepareWholeEmitSubPath(arena: Allocator, opt_emit: ?EmitLoc) error{OutOfMemo
 /// to remind the programmer to update multiple related pieces of code that
 /// are in different locations. Bump this number when adding or deleting
 /// anything from the link cache manifest.
-pub const link_hash_implementation_version = 3;
+pub const link_hash_implementation_version = 4;
 
 fn addNonIncrementalStuffToCacheManifest(comp: *Compilation, man: *Cache.Manifest) !void {
     const gpa = comp.gpa;
@@ -2369,7 +2372,7 @@ fn addNonIncrementalStuffToCacheManifest(comp: *Compilation, man: *Cache.Manifes
     defer arena_allocator.deinit();
     const arena = arena_allocator.allocator();
 
-    comptime assert(link_hash_implementation_version == 3);
+    comptime assert(link_hash_implementation_version == 4);
 
     if (comp.bin_file.options.module) |mod| {
         const main_zig_file = try mod.main_pkg.root_src_directory.join(arena, &[_][]const u8{
@@ -2474,6 +2477,7 @@ fn addNonIncrementalStuffToCacheManifest(comp: *Compilation, man: *Cache.Manifes
     man.hash.addListOfBytes(comp.bin_file.options.framework_dirs);
     man.hash.addListOfBytes(comp.bin_file.options.frameworks);
     try man.addOptionalFile(comp.bin_file.options.entitlements);
+    man.hash.addOptional(comp.bin_file.options.pagezero_size);
 
     // COFF specific stuff
     man.hash.addOptional(comp.bin_file.options.subsystem);

src/link.zig

@@ -187,6 +187,9 @@ pub const Options = struct {
     /// (Darwin) Path to entitlements file
     entitlements: ?[]const u8 = null,
 
+    /// (Darwin) size of the __PAGEZERO segment
+    pagezero_size: ?u64 = null,
+
     pub fn effectiveOutputMode(options: Options) std.builtin.OutputMode {
         return if (options.use_lld) .Obj else options.output_mode;
     }

src/link/Coff.zig

@@ -969,7 +969,7 @@ fn linkWithLLD(self: *Coff, comp: *Compilation, prog_node: *std.Progress.Node) !
         man = comp.cache_parent.obtain();
         self.base.releaseLock();
 
-        comptime assert(Compilation.link_hash_implementation_version == 3);
+        comptime assert(Compilation.link_hash_implementation_version == 4);
 
         for (self.base.options.objects) |obj| {
             _ = try man.addFile(obj.path, null);

src/link/Elf.zig

@@ -1298,7 +1298,7 @@ fn linkWithLLD(self: *Elf, comp: *Compilation, prog_node: *std.Progress.Node) !v
         // We are about to obtain this lock, so here we give other processes a chance first.
         self.base.releaseLock();
 
-        comptime assert(Compilation.link_hash_implementation_version == 3);
+        comptime assert(Compilation.link_hash_implementation_version == 4);
 
         try man.addOptionalFile(self.base.options.linker_script);
         try man.addOptionalFile(self.base.options.version_script);

src/link/MachO.zig

@@ -286,9 +286,9 @@ const default_dyld_path: [*:0]const u8 = "/usr/lib/dyld";
 const minimum_text_block_size = 64;
 pub const min_text_capacity = padToIdeal(minimum_text_block_size);
 
-/// Virtual memory offset corresponds to the size of __PAGEZERO segment and start of
+/// Default virtual memory offset corresponds to the size of __PAGEZERO segment and
-/// __TEXT segment.
+/// start of __TEXT segment.
-const pagezero_vmsize: u64 = 0x100000000;
+const default_pagezero_vmsize: u64 = 0x100000000;
 
 pub const Export = struct {
     sym_index: ?u32 = null,
@@ -536,7 +536,7 @@ pub fn flushModule(self: *MachO, comp: *Compilation, prog_node: *std.Progress.No
         // We are about to obtain this lock, so here we give other processes a chance first.
         self.base.releaseLock();
 
-        comptime assert(Compilation.link_hash_implementation_version == 3);
+        comptime assert(Compilation.link_hash_implementation_version == 4);
 
         for (self.base.options.objects) |obj| {
             _ = try man.addFile(obj.path, null);
@@ -549,6 +549,7 @@ pub fn flushModule(self: *MachO, comp: *Compilation, prog_node: *std.Progress.No
         // We can skip hashing libc and libc++ components that we are in charge of building from Zig
         // installation sources because they are always a product of the compiler version + target information.
         man.hash.add(stack_size);
+        man.hash.addOptional(self.base.options.pagezero_size);
         man.hash.addListOfBytes(self.base.options.lib_dirs);
         man.hash.addListOfBytes(self.base.options.framework_dirs);
         man.hash.addListOfBytes(self.base.options.frameworks);
@@ -928,6 +929,11 @@ pub fn flushModule(self: *MachO, comp: *Compilation, prog_node: *std.Progress.No
                     try argv.append(rpath);
                 }
 
+                if (self.base.options.pagezero_size) |pagezero_size| {
+                    try argv.append("-pagezero_size");
+                    try argv.append(try std.fmt.allocPrint(arena, "0x{x}", .{pagezero_size}));
+                }
+
                 try argv.appendSlice(positionals.items);
 
                 try argv.append("-o");
@@ -4365,14 +4371,21 @@ pub fn getDeclVAddr(self: *MachO, decl_index: Module.Decl.Index, reloc_info: Fil
 
 fn populateMissingMetadata(self: *MachO) !void {
     const cpu_arch = self.base.options.target.cpu.arch;
+    const pagezero_vmsize = self.base.options.pagezero_size orelse default_pagezero_vmsize;
+    const aligned_pagezero_vmsize = mem.alignBackwardGeneric(u64, pagezero_vmsize, self.page_size);
 
-    if (self.pagezero_segment_cmd_index == null) {
+    if (self.pagezero_segment_cmd_index == null) blk: {
+        if (aligned_pagezero_vmsize == 0) break :blk;
+        if (aligned_pagezero_vmsize != pagezero_vmsize) {
+            log.warn("requested __PAGEZERO size (0x{x}) is not page aligned", .{pagezero_vmsize});
+            log.warn("  rounding down to 0x{x}", .{aligned_pagezero_vmsize});
+        }
         self.pagezero_segment_cmd_index = @intCast(u16, self.load_commands.items.len);
         try self.load_commands.append(self.base.allocator, .{
             .segment = .{
                 .inner = .{
                     .segname = makeStaticString("__PAGEZERO"),
-                    .vmsize = pagezero_vmsize,
+                    .vmsize = aligned_pagezero_vmsize,
                     .cmdsize = @sizeOf(macho.segment_command_64),
                 },
             },
@@ -4394,7 +4407,7 @@ fn populateMissingMetadata(self: *MachO) !void {
             .segment = .{
                 .inner = .{
                     .segname = makeStaticString("__TEXT"),
-                    .vmaddr = pagezero_vmsize,
+                    .vmaddr = aligned_pagezero_vmsize,
                     .vmsize = needed_size,
                     .filesize = needed_size,
                     .maxprot = macho.PROT.READ | macho.PROT.EXEC,
@@ -4881,7 +4894,10 @@ fn populateMissingMetadata(self: *MachO) !void {
 
 fn allocateTextSegment(self: *MachO) !void {
     const seg = &self.load_commands.items[self.text_segment_cmd_index.?].segment;
-    const base_vmaddr = self.load_commands.items[self.pagezero_segment_cmd_index.?].segment.inner.vmsize;
+    const base_vmaddr = if (self.pagezero_segment_cmd_index) |index|
+        self.load_commands.items[index].segment.inner.vmsize
+    else
+        0;
     seg.inner.fileoff = 0;
     seg.inner.vmaddr = base_vmaddr;
 

src/link/Wasm.zig

@@ -2274,7 +2274,7 @@ fn linkWithLLD(self: *Wasm, comp: *Compilation, prog_node: *std.Progress.Node) !
         // We are about to obtain this lock, so here we give other processes a chance first.
         self.base.releaseLock();
 
-        comptime assert(Compilation.link_hash_implementation_version == 3);
+        comptime assert(Compilation.link_hash_implementation_version == 4);
 
         for (self.base.options.objects) |obj| {
             _ = try man.addFile(obj.path, null);

src/main.zig

@@ -447,6 +447,7 @@ const usage_build_generic =
     \\  -F[dir]                        (Darwin) add search path for frameworks
     \\  -install_name=[value]          (Darwin) add dylib's install name
     \\  --entitlements [path]          (Darwin) add path to entitlements file for embedding in code signature
+    \\  -pagezero_size [value]         (Darwin) size of the __PAGEZERO segment in hexadecimal notation
     \\  --import-memory                (WebAssembly) import memory from the environment
     \\  --import-table                 (WebAssembly) import function table from the host environment
     \\  --export-table                 (WebAssembly) export function table to the host environment
@@ -694,6 +695,7 @@ fn buildOutputType(
     var install_name: ?[]const u8 = null;
     var hash_style: link.HashStyle = .both;
     var entitlements: ?[]const u8 = null;
+    var pagezero_size: ?u64 = null;
 
     // e.g. -m3dnow or -mno-outline-atomics. They correspond to std.Target llvm cpu feature names.
     // This array is populated by zig cc frontend and then has to be converted to zig-style
@@ -908,6 +910,13 @@ fn buildOutputType(
                         install_name = args_iter.next() orelse {
                             fatal("expected parameter after {s}", .{arg});
                         };
+                    } else if (mem.eql(u8, arg, "-pagezero_size")) {
+                        const next_arg = args_iter.next() orelse {
+                            fatal("expected parameter after {s}", .{arg});
+                        };
+                        pagezero_size = std.fmt.parseUnsigned(u64, eatIntPrefix(next_arg, 16), 16) catch |err| {
+                            fatal("unable to parse '{s}': {s}", .{ arg, @errorName(err) });
+                        };
                     } else if (mem.eql(u8, arg, "-T") or mem.eql(u8, arg, "--script")) {
                         linker_script = args_iter.next() orelse {
                             fatal("expected parameter after {s}", .{arg});
@@ -1647,6 +1656,15 @@ fn buildOutputType(
                     linker_optimization = std.fmt.parseUnsigned(u8, arg["-O".len..], 10) catch |err| {
                         fatal("unable to parse '{s}': {s}", .{ arg, @errorName(err) });
                     };
+                } else if (mem.eql(u8, arg, "-pagezero_size")) {
+                    i += 1;
+                    if (i >= linker_args.items.len) {
+                        fatal("expected linker arg after '{s}'", .{arg});
+                    }
+                    const next_arg = linker_args.items[i];
+                    pagezero_size = std.fmt.parseUnsigned(u64, eatIntPrefix(next_arg, 16), 16) catch |err| {
+                        fatal("unable to parse '{s}': {s}", .{ arg, @errorName(err) });
+                    };
                 } else if (mem.eql(u8, arg, "--gc-sections")) {
                     linker_gc_sections = true;
                 } else if (mem.eql(u8, arg, "--no-gc-sections")) {
@@ -2763,6 +2781,7 @@ fn buildOutputType(
         .native_darwin_sdk = native_darwin_sdk,
         .install_name = install_name,
         .entitlements = entitlements,
+        .pagezero_size = pagezero_size,
     }) catch |err| switch (err) {
         error.LibCUnavailable => {
             const target = target_info.target;
@@ -5048,6 +5067,18 @@ pub fn cmdChangelist(
     try bw.flush();
 }
 
+fn eatIntPrefix(arg: []const u8, radix: u8) []const u8 {
+    if (arg.len > 2 and arg[0] == '0') {
+        switch (std.ascii.toLower(arg[1])) {
+            'b' => if (radix == 2) return arg[2..],
+            'o' => if (radix == 8) return arg[2..],
+            'x' => if (radix == 16) return arg[2..],
+            else => {},
+        }
+    }
+    return arg;
+}
+
 fn parseIntSuffix(arg: []const u8, prefix_len: usize) u64 {
     return std.fmt.parseUnsigned(u64, arg[prefix_len..], 0) catch |err| {
         fatal("unable to parse '{s}': {s}", .{ arg, @errorName(err) });