Unexpected EOF when constructing unit from compilation unit header

[AjayMT]

Hello all,

I would first like to thank you for your incredible work -- gimli is by far the most approachable and convenient DWARF parsing library I have found. 😄

I've been trying to parse some Mach-O files on macOS with object and gimli, and I've encountered an issue that I haven't found much documentation on. It is likely an error on my part, but I'm not sure exactly what's causing it. This is the relevant code:

let s_debug_info = parsed_file.section_by_name(".debug_info").unwrap().data();
let s_debug_abbrev = parsed_file.section_by_name(".debug_abbrev").unwrap().data();

let debug_info = gimli::DebugInfo::new(&s_debug_info, gimli::LittleEndian);
let debug_abbrev = gimli::DebugAbbrev::new(&s_debug_abbrev, gimli::LittleEndian);
let dwarf = gimli::Dwarf {
    debug_info,
    debug_abbrev,
    ..Default::default()
};

let units = dwarf.units().collect::<Vec<_>>().unwrap();
for header in units {
    let unit = match dwarf.unit(header) {
        Ok(r) => r,
        Err(err) => {
            println!("error constructing unit for header {:?}: {}", header, err);
            continue;
        }
    };

    println!("unit name {:?}", unit.name.unwrap());
}

dwarf.unit(header) seems to fail every time with an UnexpectedEof error. Perhaps this is due to how I'm loading the data with object?

I am trying to read and parse DWARF files within dSYM bundles, i.e example.dSYM/Contents/Resources/DWARF/example. Please let me know if this is not the right place to ask this question. Any advice is appreciated.

[philipc]

This is probably because dwarf.unit() tries to parse the line number program too, and you haven't provided the .debug_line section. If there's a reason you don't want to load .debug_line, then I can try to rethink the error handling in dwarf.unit() to see if we can handle this case better.

[philipc]

I see our example does the same thing, so I'll have to fix that too.

[AjayMT]

Thanks for the quick response. Unfortunately, loading debug_line does not seem to fix the error. The dwarfdump example works perfectly, so it's definitely an issue with my code. I will try loading every DWARF section.

EDIT: loading .debug_str seems to have fixed the issue, I'm not sure why.

[philipc]

Ah, .debug_str is a similar problem, we often need to use it to parse the DW_AT_name and DW_AT_compdir attributes, and you'll want it anyway if you're doing anything significant using DWARF. I've opened #406 to make it easier to load all the sections. I'm also going to see if we can make the UnexpectedEof error tell you which section the error occurred for.