.NET Core and Marshaling on Linux

Question

I got third-party native C library.so in my .NET Core 3.1 application under linux os (debian).

There was identical library.dll under Windows with .NET SDK for it. The Windows and linux C-libraries have the same methods signatures, I had check them.

With the help of dotpeek I adopt the code of sdk to linux app:

[DllImport("library.so", CharSet = CharSet.Ansi, CallingConvention = CallingConvention.StdCall)]
internal static extern uint FindEntites(out FoundResultsStruct foundResult);
...
internal ArrayList FillCollection()
{
    var innerList = new ArrayList();

    FoundResultsStruct results;
    var code = FindEntites(out results);

    IntPtr ptr = results.Value;

    if (ptr == IntPtr.Zero)
        throw new Exception();

    int num = Marshal.SizeOf(typeof(FoundEntityStruct));

    for (long index = 0; index < results.Count; ++index)
    {
        innerList.Add(new CollectionEntity(Marshal.PtrToStructure<FoundEntityStruct>(ptr))); //point of Exception
        ptr = (IntPtr)((long)ptr + (long)num);
    }

    return innerList;
}

When I run the code, it throws the exception at point of Exception at the index variable equals 0, the first element of collection:

No mapping for the Unicode character exists in the target multi-byte code page. (0x80070459)

System.ArgumentOutOfRangeException: No mapping for the Unicode character exists in the target multi-byte code page. (0x80070459)

at System.Runtime.InteropServices.Marshal.PtrToStructureHelper(IntPtr ptr, Object structure, Boolean allowValueClasses)
at System.Runtime.InteropServices.Marshal.PtrToStructureHelper(IntPtr ptr, Type structureType)
at System.Runtime.InteropServices.Marshal.PtrToStructure(IntPtr ptr, Type structureType)
at System.Runtime.InteropServices.Marshal.PtrToStructure[T](IntPtr ptr)
at MyApp.EntitesCollection.FillCollection() in C:\Projects\MyApp\EntitesCollection.cs:line 77

I suppose it cause native C-library interacts with ANSI encoding instead of UTF-8.

I had try to change the CharSet of DllImport attribute to CharSet.Ansi/Unicode/Auto - it has no effect.

Found this, looks like it's similar case, but didn't understand the solution.

Any ideas?