bad alignment cast in macho linking code

[andrewrk]

A user on IRC reported the following stack trace against zig 0.10.1:

thread 473118 panic: incorrect alignment
/private/tmp/nix-build-zig-0.10.1.drv-0/source/src/link/MachO/Object.zig:114:72: 0x1030338ab in link.MachO.Object.parse (zig2)
                    @alignCast(@alignOf(macho.nlist_64), &self.contents[symtab.symoff]),
                                                                       ^
/private/tmp/nix-build-zig-0.10.1.drv-0/source/src/link/MachO/Archive.zig:230:21: 0x1030485f7 in link.MachO.Archive.parseObject (zig2)
    try object.parse(gpa, cpu_arch);
                    ^
/private/tmp/nix-build-zig-0.10.1.drv-0/source/src/link/MachO/zld.zig:1056:55: 0x102d4af7f in link.MachO.zld.Zld.resolveSymbolsInArchives (zig2)
                const object = try archive.parseObject(gpa, cpu_arch, offsets.items[0]);
                                                      ^
/private/tmp/nix-build-zig-0.10.1.drv-0/source/src/link/MachO/zld.zig:4219:41: 0x102ad5443 in link.MachO.zld.linkWithZld (zig2)
        try zld.resolveSymbolsInArchives(&resolver);
                                        ^
/private/tmp/nix-build-zig-0.10.1.drv-0/source/src/link/MachO.zig:427:44: 0x10281667f in link.MachO.flush (zig2)
        .one_shot => return zld.linkWithZld(self, comp, prog_node),
                                           ^
/private/tmp/nix-build-zig-0.10.1.drv-0/source/src/link.zig:797:72: 0x10281345b in link.File.flush (zig2)
            .macho => return @fieldParentPtr(MachO, "base", base).flush(comp, prog_node),
                                                                       ^
/private/tmp/nix-build-zig-0.10.1.drv-0/source/src/Compilation.zig:2516:24: 0x10278ca2f in Compilation.flush (zig2)
    comp.bin_file.flush(comp, prog_node) catch |err| switch (err) {
                       ^
/private/tmp/nix-build-zig-0.10.1.drv-0/source/src/Compilation.zig:2480:27: 0x10277fb67 in Compilation.update (zig2)
            try comp.flush(main_progress_node);
                          ^
/private/tmp/nix-build-zig-0.10.1.drv-0/source/src/main.zig:3361:20: 0x102701857 in main.updateModule (zig2)
    try comp.update();
                   ^
/private/tmp/nix-build-zig-0.10.1.drv-0/source/src/main.zig:3028:17: 0x10263e09f in main.buildOutputType (zig2)
    updateModule(gpa, comp, hook) catch |err| switch (err) {
                ^
/private/tmp/nix-build-zig-0.10.1.drv-0/source/src/main.zig:230:31: 0x1025c6f6b in main.mainArgs (zig2)
        return buildOutputType(gpa, arena, args, .{ .build = .Exe });
                              ^
/private/tmp/nix-build-zig-0.10.1.drv-0/source/src/stage1.zig:56:24: 0x1025c68a3 in main (zig2)
        stage2.mainArgs(gpa, arena, args) catch unreachable;
                       ^

I examined the corresponding master branch code, and it looks to me like the problem is still there:

zig/src/link/MachO/Object.zig

Line 147 in 289e8fa

@alignCast(@alignOf(macho.nlist_64), &self.contents[symtab.symoff]),

Note that this field is marked align(1), so the alignment cast is completely unnecessary:

zig/src/link/MachO/Object.zig

Line 38 in 289e8fa

in_symtab: ?[]align(1) const macho.nlist_64 = null,

cc @kubkon

[kubkon]

Nice catch, thanks @andrewrk. There is one more issue like this in the linker codebase that I will fix together with this and #14559

[Anton-4]

Will this fix make it to 0.10.2?
I'm asking because this issue seems to be a blocker to getting zig 10 available on nix for darwin. Being able to use zig through nix is really convenient for roc contributors and also essential for when we want to release roc on nix.

[winterqt]

@Anton-4 See #14559 (comment).

@kubkon Are you still interested in backporting the patches for us? I tried a bit ago and was unsuccessful, though I suppose you can do it easier :)

[kubkon]

@winterqt Oh shoot, yeah. I was meant to do that a while back but my attention got diverted by other work. I will work on a patch for Nix shortly.

@Anton-4 There will not be a 0.10.2 patch release, and the changes will make it into 0.11.0. However, I will try to come up with a patch with @winterqt specifically for Nix in the meantime.