authorgravatar for kubkon@jakubkonka.comJakub Konka <kubkon@jakubkonka.com> 2024-02-06 11:17:00+01:00
committergravatar for kubkon@jakubkonka.comJakub Konka <kubkon@jakubkonka.com> 2024-02-06 17:38:50+01:00
logdb6addf31ac6566fc219b6dbe971c5dd96d14edc
treea155c868836b59570f7430d8a1b6de94c4ecaf00
parent7bd8b35a3dfe61e59ffea39d464e84fbcdead29a

macho: store open file descriptors in a global array

5 files changed, 68 insertions(+), 36 deletions(-)

src/link/MachO.zig+28-6
......@@ -10,6 +10,10 @@ d_sym: ?DebugSymbols = null,
1010/// Index of each input file also encodes the priority or precedence of one input file
1111/// over another.
1212files: std.MultiArrayList(File.Entry) = .{},
13/// Long-lived list of all file descriptors.
14/// We store them globally rather than per actual File so that we can re-use
15/// one file handle per every object file within an archive.
16file_handles: std.ArrayListUnmanaged(File.Handle) = .{},
1317zig_object: ?File.Index = null,
1418internal_object: ?File.Index = null,
1519objects: std.ArrayListUnmanaged(File.Index) = .{},
......@@ -315,6 +319,11 @@ pub fn deinit(self: *MachO) void {
315319 d_sym.deinit();
316320 }
317321
322 for (self.file_handles.items) |handle| {
323 handle.close();
324 }
325 self.file_handles.deinit(gpa);
326
318327 for (self.files.items(.tags), self.files.items(.data)) |tag, *data| switch (tag) {
319328 .null => {},
320329 .zig_object => data.zig_object.deinit(gpa),
......@@ -394,8 +403,6 @@ pub fn flushModule(self: *MachO, arena: Allocator, prog_node: *std.Progress.Node
394403 sub_prog_node.activate();
395404 defer sub_prog_node.end();
396405
397 const target = comp.root_mod.resolved_target.result;
398 _ = target;
399406 const directory = self.base.emit.directory;
400407 const full_out_path = try directory.join(arena, &[_][]const u8{self.base.emit.sub_path});
401408 const module_obj_path: ?[]const u8 = if (self.base.zcu_object_sub_path) |path| blk: {
......@@ -985,6 +992,8 @@ fn parseObject(self: *MachO, path: []const u8) ParseError!void {
985992
986993 const gpa = self.base.comp.gpa;
987994 const file = try std.fs.cwd().openFile(path, .{});
995 errdefer file.close();
996 const handle = try self.addFileHandle(file);
988997 const mtime: u64 = mtime: {
989998 const stat = file.stat() catch break :mtime 0;
990999 break :mtime @as(u64, @intCast(@divFloor(stat.mtime, 1_000_000_000)));
......@@ -992,7 +1001,7 @@ fn parseObject(self: *MachO, path: []const u8) ParseError!void {
9921001 const index = @as(File.Index, @intCast(try self.files.addOne(gpa)));
9931002 self.files.set(index, .{ .object = .{
9941003 .path = try gpa.dupe(u8, path),
995 .file = file,
1004 .file_handle = handle,
9961005 .mtime = mtime,
9971006 .index = index,
9981007 } });
......@@ -1020,11 +1029,12 @@ fn parseArchive(self: *MachO, lib: SystemLib, must_link: bool, fat_arch: ?fat.Ar
10201029 const gpa = self.base.comp.gpa;
10211030
10221031 const file = try std.fs.cwd().openFile(lib.path, .{});
1023 defer file.close();
1032 errdefer file.close();
1033 const handle = try self.addFileHandle(file);
10241034
10251035 var archive = Archive{};
10261036 defer archive.deinit(gpa);
1027 try archive.parse(self, lib.path, file, fat_arch);
1037 try archive.parse(self, lib.path, handle, fat_arch);
10281038
10291039 var has_parse_error = false;
10301040 for (archive.objects.items) |extracted| {
......@@ -3796,6 +3806,19 @@ pub fn getInternalObject(self: *MachO) ?*InternalObject {
37963806 return self.getFile(index).?.internal;
37973807}
37983808
3809pub fn addFileHandle(self: *MachO, file: std.fs.File) !File.HandleIndex {
3810 const gpa = self.base.comp.gpa;
3811 const index: File.HandleIndex = @intCast(self.file_handles.items.len);
3812 const fh = try self.file_handles.addOne(gpa);
3813 fh.* = file;
3814 return index;
3815}
3816
3817pub fn getFileHandle(self: MachO, index: File.HandleIndex) File.Handle {
3818 assert(index < self.file_handles.items.len);
3819 return self.file_handles.items[index];
3820}
3821
37993822pub fn addAtom(self: *MachO) error{OutOfMemory}!Atom.Index {
38003823 const index = @as(Atom.Index, @intCast(self.atoms.items.len));
38013824 const atom = try self.atoms.addOne(self.base.comp.gpa);
......@@ -4616,7 +4639,6 @@ const Cache = std.Build.Cache;
46164639const CodeSignature = @import("MachO/CodeSignature.zig");
46174640const Compilation = @import("../Compilation.zig");
46184641pub const DebugSymbols = @import("MachO/DebugSymbols.zig");
4619const Dwarf = File.Dwarf;
46204642const DwarfInfo = @import("MachO/DwarfInfo.zig");
46214643const Dylib = @import("MachO/Dylib.zig");
46224644const ExportTrieSection = synthetic.ExportTrieSection;
src/link/MachO/Archive.zig+9-7
......@@ -73,24 +73,25 @@ pub fn deinit(self: *Archive, allocator: Allocator) void {
7373 self.objects.deinit(allocator);
7474}
7575
76pub fn parse(self: *Archive, macho_file: *MachO, path: []const u8, file: std.fs.File, fat_arch: ?fat.Arch) !void {
76pub fn parse(self: *Archive, macho_file: *MachO, path: []const u8, handle_index: File.HandleIndex, fat_arch: ?fat.Arch) !void {
7777 const gpa = macho_file.base.comp.gpa;
7878
7979 var arena = std.heap.ArenaAllocator.init(gpa);
8080 defer arena.deinit();
8181
82 const handle = macho_file.getFileHandle(handle_index);
8283 const offset = if (fat_arch) |ar| ar.offset else 0;
83 const size = if (fat_arch) |ar| ar.size else (try file.stat()).size;
84 try file.seekTo(offset);
84 const size = if (fat_arch) |ar| ar.size else (try handle.stat()).size;
85 try handle.seekTo(offset);
8586
86 const reader = file.reader();
87 const reader = handle.reader();
8788 _ = try reader.readBytesNoEof(Archive.SARMAG);
8889
8990 var pos: usize = Archive.SARMAG;
9091 while (true) {
9192 if (pos >= size) break;
9293 if (!mem.isAligned(pos, 2)) {
93 try file.seekBy(1);
94 try handle.seekBy(1);
9495 pos += 1;
9596 }
9697
......@@ -118,7 +119,7 @@ pub fn parse(self: *Archive, macho_file: *MachO, path: []const u8, file: std.fs.
118119 unreachable;
119120 };
120121 defer {
121 _ = file.seekBy(hdr_size) catch {};
122 _ = handle.seekBy(hdr_size) catch {};
122123 pos += hdr_size;
123124 }
124125
......@@ -130,7 +131,7 @@ pub fn parse(self: *Archive, macho_file: *MachO, path: []const u8, file: std.fs.
130131 .offset = offset + pos,
131132 },
132133 .path = try gpa.dupe(u8, name),
133 .file = try std.fs.cwd().openFile(path, .{}),
134 .file_handle = handle_index,
134135 .index = undefined,
135136 .alive = false,
136137 .mtime = hdr.date() catch 0,
......@@ -150,5 +151,6 @@ const std = @import("std");
150151
151152const Allocator = mem.Allocator;
152153const Archive = @This();
154const File = @import("file.zig").File;
153155const MachO = @import("../MachO.zig");
154156const Object = @import("Object.zig");
src/link/MachO/Atom.zig+1-1
......@@ -58,7 +58,7 @@ pub fn getData(self: Atom, macho_file: *MachO, buffer: []u8) !void {
5858 assert(buffer.len == self.size);
5959 switch (self.getFile(macho_file)) {
6060 .internal => |x| try x.getAtomData(self, buffer),
61 .object => |x| try x.getAtomData(self, buffer),
61 .object => |x| try x.getAtomData(macho_file, self, buffer),
6262 .zig_object => |x| try x.getAtomData(macho_file, self, buffer),
6363 else => unreachable,
6464 }
src/link/MachO/Object.zig+27-22
......@@ -1,6 +1,6 @@
11archive: ?Archive = null,
22path: []const u8,
3file: std.fs.File,
3file_handle: File.HandleIndex,
44mtime: u64,
55index: File.Index,
66
......@@ -43,7 +43,6 @@ pub fn isObject(path: []const u8) !bool {
4343}
4444
4545pub fn deinit(self: *Object, allocator: Allocator) void {
46 self.file.close();
4746 if (self.archive) |*ar| allocator.free(ar.path);
4847 allocator.free(self.path);
4948 for (self.sections.items(.relocs), self.sections.items(.subsections)) |*relocs, *sub| {
......@@ -73,10 +72,11 @@ pub fn parse(self: *Object, macho_file: *MachO) !void {
7372
7473 const gpa = macho_file.base.comp.gpa;
7574 const offset = if (self.archive) |ar| ar.offset else 0;
75 const handle = macho_file.getFileHandle(self.file_handle);
7676
7777 var header_buffer: [@sizeOf(macho.mach_header_64)]u8 = undefined;
7878 {
79 const amt = try self.file.preadAll(&header_buffer, offset);
79 const amt = try handle.preadAll(&header_buffer, offset);
8080 if (amt != @sizeOf(macho.mach_header_64)) return error.InputOutput;
8181 }
8282 self.header = @as(*align(1) const macho.mach_header_64, @ptrCast(&header_buffer)).*;
......@@ -97,7 +97,7 @@ pub fn parse(self: *Object, macho_file: *MachO) !void {
9797 const lc_buffer = try gpa.alloc(u8, self.header.?.sizeofcmds);
9898 defer gpa.free(lc_buffer);
9999 {
100 const amt = try self.file.preadAll(lc_buffer, offset + @sizeOf(macho.mach_header_64));
100 const amt = try handle.preadAll(lc_buffer, offset + @sizeOf(macho.mach_header_64));
101101 if (amt != self.header.?.sizeofcmds) return error.InputOutput;
102102 }
103103
......@@ -124,14 +124,14 @@ pub fn parse(self: *Object, macho_file: *MachO) !void {
124124 const cmd = lc.cast(macho.symtab_command).?;
125125 try self.strtab.resize(gpa, cmd.strsize);
126126 {
127 const amt = try self.file.preadAll(self.strtab.items, cmd.stroff + offset);
127 const amt = try handle.preadAll(self.strtab.items, cmd.stroff + offset);
128128 if (amt != self.strtab.items.len) return error.InputOutput;
129129 }
130130
131131 const symtab_buffer = try gpa.alloc(u8, cmd.nsyms * @sizeOf(macho.nlist_64));
132132 defer gpa.free(symtab_buffer);
133133 {
134 const amt = try self.file.preadAll(symtab_buffer, cmd.symoff + offset);
134 const amt = try handle.preadAll(symtab_buffer, cmd.symoff + offset);
135135 if (amt != symtab_buffer.len) return error.InputOutput;
136136 }
137137 const symtab = @as([*]align(1) const macho.nlist_64, @ptrCast(symtab_buffer.ptr))[0..cmd.nsyms];
......@@ -149,7 +149,7 @@ pub fn parse(self: *Object, macho_file: *MachO) !void {
149149 const buffer = try gpa.alloc(u8, cmd.datasize);
150150 defer gpa.free(buffer);
151151 {
152 const amt = try self.file.preadAll(buffer, offset + cmd.dataoff);
152 const amt = try handle.preadAll(buffer, offset + cmd.dataoff);
153153 if (amt != buffer.len) return error.InputOutput;
154154 }
155155 const ndice = @divExact(cmd.datasize, @sizeOf(macho.data_in_code_entry));
......@@ -697,7 +697,7 @@ fn initEhFrameRecords(self: *Object, sect_id: u8, macho_file: *MachO) !void {
697697 const relocs = slice.items(.relocs)[sect_id];
698698
699699 // TODO: read into buffer directly
700 const data = try self.getSectionData(gpa, sect_id);
700 const data = try self.getSectionData(sect_id, macho_file);
701701 defer gpa.free(data);
702702
703703 try self.eh_frame_data.ensureTotalCapacityPrecise(gpa, data.len);
......@@ -800,7 +800,7 @@ fn initUnwindRecords(self: *Object, sect_id: u8, macho_file: *MachO) !void {
800800 };
801801
802802 const gpa = macho_file.base.comp.gpa;
803 const data = try self.getSectionData(gpa, sect_id);
803 const data = try self.getSectionData(sect_id, macho_file);
804804 defer gpa.free(data);
805805 const nrecs = @divExact(data.len, @sizeOf(macho.compact_unwind_entry));
806806 const recs = @as([*]align(1) const macho.compact_unwind_entry, @ptrCast(data.ptr))[0..nrecs];
......@@ -1019,11 +1019,11 @@ fn initDwarfInfo(self: *Object, macho_file: *MachO) !void {
10191019
10201020 if (debug_info_index == null or debug_abbrev_index == null) return;
10211021
1022 const debug_info = try self.getSectionData(gpa, @intCast(debug_info_index.?));
1022 const debug_info = try self.getSectionData(@intCast(debug_info_index.?), macho_file);
10231023 defer gpa.free(debug_info);
1024 const debug_abbrev = try self.getSectionData(gpa, @intCast(debug_abbrev_index.?));
1024 const debug_abbrev = try self.getSectionData(@intCast(debug_abbrev_index.?), macho_file);
10251025 defer gpa.free(debug_abbrev);
1026 const debug_str = if (debug_str_index) |index| try self.getSectionData(gpa, @intCast(index)) else &[0]u8{};
1026 const debug_str = if (debug_str_index) |index| try self.getSectionData(@intCast(index), macho_file) else &[0]u8{};
10271027 defer gpa.free(debug_str);
10281028
10291029 var dwarf_info = DwarfInfo{};
......@@ -1589,25 +1589,28 @@ pub fn writeStabs(self: *const Object, macho_file: *MachO, ctx: anytype) error{O
15891589 }
15901590}
15911591
1592fn getSectionData(self: *const Object, allocator: Allocator, index: u32) ![]u8 {
1592fn getSectionData(self: *const Object, index: u32, macho_file: *MachO) ![]u8 {
1593 const gpa = macho_file.base.comp.gpa;
15931594 const slice = self.sections.slice();
15941595 assert(index < slice.items(.header).len);
15951596 const sect = slice.items(.header)[index];
1597 const handle = macho_file.getFileHandle(self.file_handle);
15961598 const offset = if (self.archive) |ar| ar.offset else 0;
15971599 const size = math.cast(usize, sect.size) orelse return error.Overflow;
1598 const buffer = try allocator.alloc(u8, size);
1599 errdefer allocator.free(buffer);
1600 const amt = try self.file.preadAll(buffer, sect.offset + offset);
1600 const buffer = try gpa.alloc(u8, size);
1601 errdefer gpa.free(buffer);
1602 const amt = try handle.preadAll(buffer, sect.offset + offset);
16011603 if (amt != buffer.len) return error.InputOutput;
16021604 return buffer;
16031605}
16041606
1605pub fn getAtomData(self: *const Object, atom: Atom, buffer: []u8) !void {
1607pub fn getAtomData(self: *const Object, macho_file: *MachO, atom: Atom, buffer: []u8) !void {
16061608 assert(buffer.len == atom.size);
16071609 const slice = self.sections.slice();
1610 const handle = macho_file.getFileHandle(self.file_handle);
16081611 const offset = if (self.archive) |ar| ar.offset else 0;
16091612 const sect = slice.items(.header)[atom.n_sect];
1610 const amt = try self.file.preadAll(buffer, sect.offset + offset + atom.off);
1613 const amt = try handle.preadAll(buffer, sect.offset + offset + atom.off);
16111614 if (amt != buffer.len) return error.InputOutput;
16121615}
16131616
......@@ -1885,16 +1888,17 @@ const x86_64 = struct {
18851888 ) !void {
18861889 const gpa = macho_file.base.comp.gpa;
18871890
1891 const handle = macho_file.getFileHandle(self.file_handle);
18881892 const offset = if (self.archive) |ar| ar.offset else 0;
18891893 const relocs_buffer = try gpa.alloc(u8, sect.nreloc * @sizeOf(macho.relocation_info));
18901894 defer gpa.free(relocs_buffer);
18911895 {
1892 const amt = try self.file.preadAll(relocs_buffer, sect.reloff + offset);
1896 const amt = try handle.preadAll(relocs_buffer, sect.reloff + offset);
18931897 if (amt != relocs_buffer.len) return error.InputOutput;
18941898 }
18951899 const relocs = @as([*]align(1) const macho.relocation_info, @ptrCast(relocs_buffer.ptr))[0..sect.nreloc];
18961900
1897 const code = try self.getSectionData(gpa, @intCast(n_sect));
1901 const code = try self.getSectionData(@intCast(n_sect), macho_file);
18981902 defer gpa.free(code);
18991903
19001904 try out.ensureTotalCapacityPrecise(gpa, relocs.len);
......@@ -2047,16 +2051,17 @@ const aarch64 = struct {
20472051 ) !void {
20482052 const gpa = macho_file.base.comp.gpa;
20492053
2054 const handle = macho_file.getFileHandle(self.file_handle);
20502055 const offset = if (self.archive) |ar| ar.offset else 0;
20512056 const relocs_buffer = try gpa.alloc(u8, sect.nreloc * @sizeOf(macho.relocation_info));
20522057 defer gpa.free(relocs_buffer);
20532058 {
2054 const amt = try self.file.preadAll(relocs_buffer, sect.reloff + offset);
2059 const amt = try handle.preadAll(relocs_buffer, sect.reloff + offset);
20552060 if (amt != relocs_buffer.len) return error.InputOutput;
20562061 }
20572062 const relocs = @as([*]align(1) const macho.relocation_info, @ptrCast(relocs_buffer.ptr))[0..sect.nreloc];
20582063
2059 const code = try self.getSectionData(gpa, @intCast(n_sect));
2064 const code = try self.getSectionData(@intCast(n_sect), macho_file);
20602065 defer gpa.free(code);
20612066
20622067 try out.ensureTotalCapacityPrecise(gpa, relocs.len);
src/link/MachO/file.zig+3
......@@ -105,6 +105,9 @@ pub const File = union(enum) {
105105 object: Object,
106106 dylib: Dylib,
107107 };
108
109 pub const Handle = std.fs.File;
110 pub const HandleIndex = Index;
108111};
109112
110113const macho = std.macho;