attempt to print stack trace addresses

[jfo]

I'm just experimenting, but I'm at

https://www.unix.com/man-page/osx/3/backtrace/
and
https://searchfox.org/mozilla-central/source/widget/cocoa/nsCocoaDebugUtils.h#11
and
rust-lang/rust#24346

[jfo]

when run with our friend:

// err.zig

const std = @import("std");

pub fn main() !void {
    const allocator = std.debug.global_allocator;

    const args = try std.os.argsAlloc(allocator);
    defer std.os.argsFree(allocator, args);

    const count = try parseFile(allocator, args[1]);

    if (count < 10) return error.NotEnoughItems;
}

fn parseFile(allocator: &std.mem.Allocator, file_path: []const u8) !usize {
    const contents = std.io.readFileAlloc(allocator, file_path) catch return error.UnableToReadFile;
    defer allocator.free(contents);

    return contents.len;
}

looks something like

0   err                                 0x000000010514ce73 openSelfDebugInfo + 147
1   err                                 0x000000010514c346 getSelfDebugInfo + 150
2   err                                 0x000000010514bc79 dumpCurrentStackTrace + 169
3   err                                 0x000000010514b634 panic.6 + 212
4   err                                 0x000000010514b46b panic + 91
5   err                                 0x000000010515b13e main.0 + 446
6   err                                 0x000000010515ae59 callMain + 73
7   err                                 0x000000010515add9 callMainWithArgs + 361
8   err                                 0x000000010515ac64 main + 36
9   libdyld.dylib                       0x00007fff7af96115 start + 1
10  ???                                 0x0000000000000001 0x0 + 1

[jfo]

also re: ziglang#434 (comment) see: https://www.unix.com/man-page/osx/3/dladdr/

[andrewrk]

We don't want to introduce a dependency on .h files and libc in the standard library.

What we can do is read the source for the code you're calling into and port it into zig

[jfo]

Yes, I know that about the libc dep, hence the "just experimenting"

But, is this headed in the right direction?

[andrewrk]

I think that the existing dwarf code can actually support macos, but it incorrectly looks for it in its own binary, whereas in reality you have to find the object files and look there. I'd start with trying to figure out, perhaps by reading the implementation of the code you just called into, how to find the actual dwarf info

[jfo]

❤️ SO

also

[andrewrk]

Yes I think these are the keys to doing it. cc @bnoordhuis are you still working on this?

[bnoordhuis]

Yes, but using the symtab segment from the image, not external dSYM data. It's less precise but at least it's always available. I'll try to PR it in the next few days.