macho: report special symbols if undefined

Special symbols include explictly force undefined symbols passed via -u
flag, missing entry point symbol, missing 'dyld_stub_binder' symbol, or
missing '_objc_msgsend' symbol.
This commit is contained in:
Jakub Konka 2024-10-09 17:06:34 +02:00
parent 22661f3d67
commit 90aa68cbfc
3 changed files with 162 additions and 88 deletions

View File

@ -27,7 +27,7 @@ sections: std.MultiArrayList(Section) = .{},
resolver: SymbolResolver = .{}, resolver: SymbolResolver = .{},
/// This table will be populated after `scanRelocs` has run. /// This table will be populated after `scanRelocs` has run.
/// Key is symbol index. /// Key is symbol index.
undefs: std.AutoArrayHashMapUnmanaged(SymbolResolver.Index, std.ArrayListUnmanaged(Ref)) = .empty, undefs: std.AutoArrayHashMapUnmanaged(SymbolResolver.Index, UndefRefs) = .empty,
undefs_mutex: std.Thread.Mutex = .{}, undefs_mutex: std.Thread.Mutex = .{},
dupes: std.AutoArrayHashMapUnmanaged(SymbolResolver.Index, std.ArrayListUnmanaged(File.Index)) = .empty, dupes: std.AutoArrayHashMapUnmanaged(SymbolResolver.Index, std.ArrayListUnmanaged(File.Index)) = .empty,
dupes_mutex: std.Thread.Mutex = .{}, dupes_mutex: std.Thread.Mutex = .{},
@ -1470,6 +1470,9 @@ fn scanRelocs(self: *MachO) !void {
if (self.has_errors.swap(false, .seq_cst)) return error.FlushFailure; if (self.has_errors.swap(false, .seq_cst)) return error.FlushFailure;
if (self.getInternalObject()) |obj| {
try obj.checkUndefs(self);
}
try self.reportUndefs(); try self.reportUndefs();
if (self.getZigObject()) |zo| { if (self.getZigObject()) |zo| {
@ -1530,29 +1533,43 @@ fn reportUndefs(self: *MachO) !void {
} }
}.lessThan; }.lessThan;
for (self.undefs.values()) |*refs| { for (self.undefs.values()) |*undefs| switch (undefs.*) {
mem.sort(Ref, refs.items, {}, refLessThan); .refs => |refs| mem.sort(Ref, refs.items, {}, refLessThan),
} else => {},
};
for (keys.items) |key| { for (keys.items) |key| {
const undef_sym = self.resolver.keys.items[key - 1]; const undef_sym = self.resolver.keys.items[key - 1];
const notes = self.undefs.get(key).?; const notes = self.undefs.get(key).?;
const nnotes = @min(notes.items.len, max_notes) + @intFromBool(notes.items.len > max_notes); const nnotes = nnotes: {
const nnotes = switch (notes) {
.refs => |refs| refs.items.len,
else => 1,
};
break :nnotes @min(nnotes, max_notes) + @intFromBool(nnotes > max_notes);
};
var err = try self.base.addErrorWithNotes(nnotes); var err = try self.base.addErrorWithNotes(nnotes);
try err.addMsg("undefined symbol: {s}", .{undef_sym.getName(self)}); try err.addMsg("undefined symbol: {s}", .{undef_sym.getName(self)});
var inote: usize = 0; switch (notes) {
while (inote < @min(notes.items.len, max_notes)) : (inote += 1) { .force_undefined => try err.addNote("referenced with linker flag -u", .{}),
const note = notes.items[inote]; .entry => try err.addNote("referenced with linker flag -e", .{}),
const file = self.getFile(note.file).?; .dyld_stub_binder, .objc_msgsend => try err.addNote("referenced implicitly", .{}),
const atom = note.getAtom(self).?; .refs => |refs| {
try err.addNote("referenced by {}:{s}", .{ file.fmtPath(), atom.getName(self) }); var inote: usize = 0;
} while (inote < @min(refs.items.len, max_notes)) : (inote += 1) {
const ref = refs.items[inote];
const file = self.getFile(ref.file).?;
const atom = ref.getAtom(self).?;
try err.addNote("referenced by {}:{s}", .{ file.fmtPath(), atom.getName(self) });
}
if (notes.items.len > max_notes) { if (refs.items.len > max_notes) {
const remaining = notes.items.len - max_notes; const remaining = refs.items.len - max_notes;
try err.addNote("referenced {d} more times", .{remaining}); try err.addNote("referenced {d} more times", .{remaining});
}
},
} }
} }
@ -4584,78 +4601,20 @@ pub const String = struct {
len: u32 = 0, len: u32 = 0,
}; };
const MachO = @This(); pub const UndefRefs = union(enum) {
force_undefined,
entry,
dyld_stub_binder,
objc_msgsend,
refs: std.ArrayListUnmanaged(Ref),
const std = @import("std"); pub fn deinit(self: *UndefRefs, allocator: Allocator) void {
const build_options = @import("build_options"); switch (self.*) {
const builtin = @import("builtin"); .refs => |*refs| refs.deinit(allocator),
const assert = std.debug.assert; else => {},
const fs = std.fs; }
const log = std.log.scoped(.link); }
const state_log = std.log.scoped(.link_state); };
const macho = std.macho;
const math = std.math;
const mem = std.mem;
const meta = std.meta;
const aarch64 = @import("../arch/aarch64/bits.zig");
const bind = @import("MachO/dyld_info/bind.zig");
const calcUuid = @import("MachO/uuid.zig").calcUuid;
const codegen = @import("../codegen.zig");
const dead_strip = @import("MachO/dead_strip.zig");
const eh_frame = @import("MachO/eh_frame.zig");
const fat = @import("MachO/fat.zig");
const link = @import("../link.zig");
const load_commands = @import("MachO/load_commands.zig");
const relocatable = @import("MachO/relocatable.zig");
const tapi = @import("tapi.zig");
const target_util = @import("../target.zig");
const trace = @import("../tracy.zig").trace;
const synthetic = @import("MachO/synthetic.zig");
const Air = @import("../Air.zig");
const Alignment = Atom.Alignment;
const Allocator = mem.Allocator;
const Archive = @import("MachO/Archive.zig");
pub const Atom = @import("MachO/Atom.zig");
const AtomicBool = std.atomic.Value(bool);
const Bind = bind.Bind;
const Cache = std.Build.Cache;
const Path = Cache.Path;
const CodeSignature = @import("MachO/CodeSignature.zig");
const Compilation = @import("../Compilation.zig");
const DataInCode = synthetic.DataInCode;
pub const DebugSymbols = @import("MachO/DebugSymbols.zig");
const Dylib = @import("MachO/Dylib.zig");
const ExportTrie = @import("MachO/dyld_info/Trie.zig");
const File = @import("MachO/file.zig").File;
const GotSection = synthetic.GotSection;
const Hash = std.hash.Wyhash;
const Indsymtab = synthetic.Indsymtab;
const InternalObject = @import("MachO/InternalObject.zig");
const ObjcStubsSection = synthetic.ObjcStubsSection;
const Object = @import("MachO/Object.zig");
const LazyBind = bind.LazyBind;
const LaSymbolPtrSection = synthetic.LaSymbolPtrSection;
const Liveness = @import("../Liveness.zig");
const LlvmObject = @import("../codegen/llvm.zig").Object;
const Md5 = std.crypto.hash.Md5;
const Zcu = @import("../Zcu.zig");
const InternPool = @import("../InternPool.zig");
const Rebase = @import("MachO/dyld_info/Rebase.zig");
pub const Relocation = @import("MachO/Relocation.zig");
const StringTable = @import("StringTable.zig");
const StubsSection = synthetic.StubsSection;
const StubsHelperSection = synthetic.StubsHelperSection;
const Symbol = @import("MachO/Symbol.zig");
const Thunk = @import("MachO/Thunk.zig");
const TlvPtrSection = synthetic.TlvPtrSection;
const Value = @import("../Value.zig");
const UnwindInfo = @import("MachO/UnwindInfo.zig");
const WaitGroup = std.Thread.WaitGroup;
const WeakBind = bind.WeakBind;
const ZigObject = @import("MachO/ZigObject.zig");
const dev = @import("../dev.zig");
pub const MachError = error{ pub const MachError = error{
/// Not enough permissions held to perform the requested kernel /// Not enough permissions held to perform the requested kernel
@ -5392,3 +5351,76 @@ const max_distance = (1 << (jump_bits - 1));
/// mold uses 5MiB margin, while ld64 uses 4MiB margin. We will follow mold /// mold uses 5MiB margin, while ld64 uses 4MiB margin. We will follow mold
/// and assume margin to be 5MiB. /// and assume margin to be 5MiB.
const max_allowed_distance = max_distance - 0x500_000; const max_allowed_distance = max_distance - 0x500_000;
const MachO = @This();
const std = @import("std");
const build_options = @import("build_options");
const builtin = @import("builtin");
const assert = std.debug.assert;
const fs = std.fs;
const log = std.log.scoped(.link);
const state_log = std.log.scoped(.link_state);
const macho = std.macho;
const math = std.math;
const mem = std.mem;
const meta = std.meta;
const aarch64 = @import("../arch/aarch64/bits.zig");
const bind = @import("MachO/dyld_info/bind.zig");
const calcUuid = @import("MachO/uuid.zig").calcUuid;
const codegen = @import("../codegen.zig");
const dead_strip = @import("MachO/dead_strip.zig");
const eh_frame = @import("MachO/eh_frame.zig");
const fat = @import("MachO/fat.zig");
const link = @import("../link.zig");
const load_commands = @import("MachO/load_commands.zig");
const relocatable = @import("MachO/relocatable.zig");
const tapi = @import("tapi.zig");
const target_util = @import("../target.zig");
const trace = @import("../tracy.zig").trace;
const synthetic = @import("MachO/synthetic.zig");
const Air = @import("../Air.zig");
const Alignment = Atom.Alignment;
const Allocator = mem.Allocator;
const Archive = @import("MachO/Archive.zig");
pub const Atom = @import("MachO/Atom.zig");
const AtomicBool = std.atomic.Value(bool);
const Bind = bind.Bind;
const Cache = std.Build.Cache;
const Path = Cache.Path;
const CodeSignature = @import("MachO/CodeSignature.zig");
const Compilation = @import("../Compilation.zig");
const DataInCode = synthetic.DataInCode;
pub const DebugSymbols = @import("MachO/DebugSymbols.zig");
const Dylib = @import("MachO/Dylib.zig");
const ExportTrie = @import("MachO/dyld_info/Trie.zig");
const File = @import("MachO/file.zig").File;
const GotSection = synthetic.GotSection;
const Hash = std.hash.Wyhash;
const Indsymtab = synthetic.Indsymtab;
const InternalObject = @import("MachO/InternalObject.zig");
const ObjcStubsSection = synthetic.ObjcStubsSection;
const Object = @import("MachO/Object.zig");
const LazyBind = bind.LazyBind;
const LaSymbolPtrSection = synthetic.LaSymbolPtrSection;
const Liveness = @import("../Liveness.zig");
const LlvmObject = @import("../codegen/llvm.zig").Object;
const Md5 = std.crypto.hash.Md5;
const Zcu = @import("../Zcu.zig");
const InternPool = @import("../InternPool.zig");
const Rebase = @import("MachO/dyld_info/Rebase.zig");
pub const Relocation = @import("MachO/Relocation.zig");
const StringTable = @import("StringTable.zig");
const StubsSection = synthetic.StubsSection;
const StubsHelperSection = synthetic.StubsHelperSection;
const Symbol = @import("MachO/Symbol.zig");
const Thunk = @import("MachO/Thunk.zig");
const TlvPtrSection = synthetic.TlvPtrSection;
const Value = @import("../Value.zig");
const UnwindInfo = @import("MachO/UnwindInfo.zig");
const WaitGroup = std.Thread.WaitGroup;
const WeakBind = bind.WeakBind;
const ZigObject = @import("MachO/ZigObject.zig");
const dev = @import("../dev.zig");

View File

@ -560,9 +560,9 @@ fn reportUndefSymbol(self: Atom, rel: Relocation, macho_file: *MachO) !bool {
const gpa = macho_file.base.comp.gpa; const gpa = macho_file.base.comp.gpa;
const gop = try macho_file.undefs.getOrPut(gpa, file.getGlobals()[rel.target]); const gop = try macho_file.undefs.getOrPut(gpa, file.getGlobals()[rel.target]);
if (!gop.found_existing) { if (!gop.found_existing) {
gop.value_ptr.* = .{}; gop.value_ptr.* = .{ .refs = .{} };
} }
try gop.value_ptr.append(gpa, .{ .index = self.atom_index, .file = self.file }); try gop.value_ptr.refs.append(gpa, .{ .index = self.atom_index, .file = self.file });
return true; return true;
} }

View File

@ -507,6 +507,41 @@ pub fn scanRelocs(self: *InternalObject, macho_file: *MachO) void {
} }
} }
pub fn checkUndefs(self: InternalObject, macho_file: *MachO) !void {
const addUndef = struct {
fn addUndef(mf: *MachO, index: MachO.SymbolResolver.Index, tag: anytype) !void {
const gpa = mf.base.comp.gpa;
mf.undefs_mutex.lock();
defer mf.undefs_mutex.unlock();
const gop = try mf.undefs.getOrPut(gpa, index);
if (!gop.found_existing) {
gop.value_ptr.* = tag;
}
}
}.addUndef;
for (self.force_undefined.items) |index| {
const ref = self.getSymbolRef(index, macho_file);
if (ref.getFile(macho_file) == null) {
try addUndef(macho_file, self.globals.items[index], .force_undefined);
}
}
if (self.getEntryRef(macho_file)) |ref| {
if (ref.getFile(macho_file) == null) {
try addUndef(macho_file, self.globals.items[self.entry_index.?], .entry);
}
}
if (self.getDyldStubBinderRef(macho_file)) |ref| {
if (ref.getFile(macho_file) == null and macho_file.stubs.symbols.items.len > 0) {
try addUndef(macho_file, self.globals.items[self.dyld_stub_binder_index.?], .dyld_stub_binder);
}
}
if (self.getObjcMsgSendRef(macho_file)) |ref| {
if (ref.getFile(macho_file) == null and self.needsObjcMsgsendSymbol()) {
try addUndef(macho_file, self.globals.items[self.objc_msg_send_index.?], .objc_msgsend);
}
}
}
pub fn allocateSyntheticSymbols(self: *InternalObject, macho_file: *MachO) void { pub fn allocateSyntheticSymbols(self: *InternalObject, macho_file: *MachO) void {
const text_seg = macho_file.getTextSegment(); const text_seg = macho_file.getTextSegment();
@ -791,6 +826,13 @@ pub fn setSymbolExtra(self: *InternalObject, index: u32, extra: Symbol.Extra) vo
} }
} }
fn needsObjcMsgsendSymbol(self: InternalObject) bool {
for (self.sections.items(.extra)) |extra| {
if (extra.is_objc_methname or extra.is_objc_selref) return true;
}
return false;
}
const FormatContext = struct { const FormatContext = struct {
self: *InternalObject, self: *InternalObject,
macho_file: *MachO, macho_file: *MachO,