COFF header is changed #620

New issue

Closed

opened 2019-10-15 00:58:45 -05:00 by nxtn · 4 comments

nxtn commented

2019-10-15 00:58:45 -05:00

(Migrated from github.com)

Steps to reproduce

using System.Diagnostics;
using System.IO;
using System.Reflection.PortableExecutable;
using Mono.Cecil;

public class Program
{
    public static void Main()
    {
        const string Path = "executable.dll";

        using (var reader = new PEReader(new FileStream(Path, FileMode.Open)))
        {
            Debug.Assert(reader.PEHeaders.CoffHeader.Characteristics ==
                (Characteristics.ExecutableImage | Characteristics.LargeAddressAware));
        }

        using (var module = ModuleDefinition.ReadModule(Path, new ReaderParameters { ReadWrite = true }))
        {
            module.Write();
        }

        using (var reader = new PEReader(new FileStream(Path, FileMode.Open)))
        {
            Debug.Assert(reader.PEHeaders.CoffHeader.Characteristics ==
                (Characteristics.ExecutableImage | Characteristics.Bit32Machine));
        }
    }
}

Steps to reproduce ```cs using System.Diagnostics; using System.IO; using System.Reflection.PortableExecutable; using Mono.Cecil; public class Program { public static void Main() { const string Path = "executable.dll"; using (var reader = new PEReader(new FileStream(Path, FileMode.Open))) { Debug.Assert(reader.PEHeaders.CoffHeader.Characteristics == (Characteristics.ExecutableImage | Characteristics.LargeAddressAware)); } using (var module = ModuleDefinition.ReadModule(Path, new ReaderParameters { ReadWrite = true })) { module.Write(); } using (var reader = new PEReader(new FileStream(Path, FileMode.Open))) { Debug.Assert(reader.PEHeaders.CoffHeader.Characteristics == (Characteristics.ExecutableImage | Characteristics.Bit32Machine)); } } } ```

👍 1

jbevain commented

2019-10-15 13:32:01 -05:00

(Migrated from github.com)

Thanks for reporting this.

Cecil is recomputing the headers, it might produce a difference after round tripping.

The issue is that we're losing the LargeAddressAware flag here?

Thanks for reporting this. Cecil is recomputing the headers, it might produce a difference after round tripping. The issue is that we're losing the LargeAddressAware flag here?

nxtn commented

2019-10-15 21:06:18 -05:00

(Migrated from github.com)