Fix IcedAssembler immediate overflow and unwrap invoke exceptions

- TryBind wraps Convert.ChangeType in TryChangeType so an immediate that
  overflows a candidate parameter type returns false (letting a wider overload
  be tried) instead of throwing OverflowException out of assembly. Verified:
  "mov eax, 4294967295" and "mov eax, -2147483649" no longer crash.
- Immediate now carries a boxed long OR ulong; TryParseImmediate parses decimal
  values above long.MaxValue via a ulong fallback, and hex via ulong. Previously
  such literals were rejected at parse time.
- Unwrap TargetInvocationException from method.Invoke so callers see the real
  Iced failure, not the reflection wrapper.

Tests: 231 passing, 4 skipped.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

.gitignore

@@ -10,3 +10,7 @@ reference/
 # Scratch
 *.tmp
 *.log
+
+# Test run artifacts
+WhiteMagicTest/TestResults/
+**/TestResults/

WhiteMagic/Assembly/IcedAssembler.cs

@@ -0,0 +1,333 @@
+using System.Collections.Generic;
+using System.Globalization;
+using System.Reflection;
+using Iced.Intel;
+
+namespace WhiteMagic.Assembly;
+
+/// <summary>
+/// Optional <see cref="IAssembler"/> backend that assembles arbitrary x86/x64 mnemonic
+/// text to machine code using the Iced library, and provides full instruction-boundary
+/// decoding for detour prologue validation.
+/// </summary>
+/// <remarks>
+/// <para>Iced ships a fluent code assembler (typed method calls) and a decoder, but no
+/// text parser. This class bridges Intel-syntax text onto Iced's fluent
+/// <see cref="Assembler"/> by reflection: each line's mnemonic selects the matching
+/// <see cref="Assembler"/> method and its operands are bound to registers, immediates, or
+/// labels. Register and immediate operands and label-relative branches are supported;
+/// memory operands (<c>[reg+disp]</c>) are not — a caller needing those should emit bytes
+/// directly.</para>
+/// <para>This backend is entirely optional. Constructing it is the only thing that pulls
+/// Iced into a behavioral path; the default <see cref="StubAssembler"/> never references it.</para>
+/// </remarks>
+public sealed class IcedAssembler : IAssembler
+{
+    private const int DefaultBitness = 64;
+
+    private readonly int _bitness;
+
+    // Lowercased register name -> boxed AssemblerRegisterNN value, built once from
+    // Iced's AssemblerRegisters. Enables binding a text operand like "esp" to a typed
+    // fluent-API register argument.
+    private static readonly Dictionary<string, object> Registers = BuildRegisterMap();
+
+    /// <summary>Creates an assembler for the given bitness (32 or 64).</summary>
+    /// <param name="bitness">32 for x86, 64 for x64. Defaults to 64.</param>
+    public IcedAssembler(int bitness = DefaultBitness)
+    {
+        if (bitness != 32 && bitness != 64)
+            throw new ArgumentOutOfRangeException(nameof(bitness), "Bitness must be 32 or 64.");
+
+        _bitness = bitness;
+    }
+
+    /// <inheritdoc />
+    public byte[] Assemble(string assemblyText, ulong origin = 0)
+    {
+        ArgumentNullException.ThrowIfNull(assemblyText);
+
+        var assembler = new Assembler(_bitness);
+
+        List<(string Mnemonic, string[] Operands)> lines = Tokenize(assemblyText, out var labelNames);
+
+        // Pre-create every label so a forward branch can reference it before its definition.
+        var labels = new Dictionary<string, Label>(StringComparer.OrdinalIgnoreCase);
+        foreach (string name in labelNames)
+            labels[name] = assembler.CreateLabel(name);
+
+        foreach ((string mnemonic, string[] operands) in lines)
+        {
+            // A pure label definition (e.g. "loop:") marks the current position.
+            if (mnemonic.EndsWith(':'))
+            {
+                Label label = labels[mnemonic[..^1]];
+                assembler.Label(ref label);
+                continue;
+            }
+
+            EmitInstruction(assembler, mnemonic, operands, labels);
+        }
+
+        var writer = new ByteListCodeWriter();
+        assembler.Assemble(writer, origin);
+        return writer.Bytes.ToArray();
+    }
+
+    /// <summary>
+    /// Computes the number of whole prologue-instruction bytes that must be preserved for
+    /// a splice of <paramref name="requiredBytes"/> bytes, decoding arbitrary instructions
+    /// (not just the common prologue shapes the built-in decoder covers). Matches the
+    /// <c>PrologueLengthResolver</c> delegate so it can be assigned to
+    /// <see cref="WhiteMagic.Hooking.DetourManager.PrologueLengthResolver"/>.
+    /// </summary>
+    /// <exception cref="InvalidOperationException">A prologue byte sequence does not decode
+    /// to a valid instruction.</exception>
+    public int GetPrologueLength(byte[] prologue, int requiredBytes, bool is64Bit)
+    {
+        ArgumentNullException.ThrowIfNull(prologue);
+
+        var reader = new ByteArrayCodeReader(prologue);
+        var decoder = Decoder.Create(is64Bit ? 64 : 32, reader);
+
+        int total = 0;
+        while (total < requiredBytes)
+        {
+            decoder.Decode(out Instruction instruction);
+            if (instruction.IsInvalid)
+            {
+                throw new InvalidOperationException(
+                    "The target prologue contains a byte sequence that does not decode to a valid instruction.");
+            }
+
+            total += instruction.Length;
+        }
+
+        return total;
+    }
+
+    private void EmitInstruction(
+        Assembler assembler,
+        string mnemonic,
+        string[] operandText,
+        Dictionary<string, Label> labels)
+    {
+        object?[] operands = new object?[operandText.Length];
+        for (int i = 0; i < operandText.Length; i++)
+            operands[i] = ParseOperand(operandText[i], labels);
+
+        // Find the fluent Assembler method whose name equals the mnemonic and whose
+        // parameters bind to the parsed operands.
+        foreach (MethodInfo method in typeof(Assembler).GetMethods(BindingFlags.Public | BindingFlags.Instance))
+        {
+            if (!string.Equals(method.Name, mnemonic, StringComparison.OrdinalIgnoreCase))
+                continue;
+
+            ParameterInfo[] parameters = method.GetParameters();
+            if (parameters.Length != operands.Length)
+                continue;
+
+            if (TryBind(parameters, operands, out object?[]? boundArgs))
+            {
+                try
+                {
+                    method.Invoke(assembler, boundArgs);
+                }
+                catch (TargetInvocationException ex) when (ex.InnerException is not null)
+                {
+                    // Surface the real Iced failure rather than the reflection wrapper.
+                    throw ex.InnerException;
+                }
+
+                return;
+            }
+        }
+
+        throw new NotSupportedException(
+            $"Cannot assemble '{mnemonic}{(operandText.Length > 0 ? " " + string.Join(", ", operandText) : "")}': " +
+            "no matching Iced assembler overload for the given operands (registers, immediates and " +
+            "labels are supported; memory operands are not).");
+    }
+
+    private static bool TryBind(ParameterInfo[] parameters, object?[] operands, out object?[]? boundArgs)
+    {
+        var args = new object?[parameters.Length];
+        for (int i = 0; i < parameters.Length; i++)
+        {
+            Type paramType = parameters[i].ParameterType;
+            object? operand = operands[i];
+
+            switch (operand)
+            {
+                case Immediate imm when IsNumeric(paramType):
+                    // An immediate that overflows this parameter's type means this overload
+                    // is the wrong width; return false so a wider overload can be tried
+                    // instead of crashing the whole assembly.
+                    if (!TryChangeType(imm.Value, paramType, out object? converted))
+                    {
+                        boundArgs = null;
+                        return false;
+                    }
+
+                    args[i] = converted;
+                    break;
+
+                case not null when paramType.IsInstanceOfType(operand):
+                    args[i] = operand;
+                    break;
+
+                default:
+                    boundArgs = null;
+                    return false;
+            }
+        }
+
+        boundArgs = args;
+        return true;
+    }
+
+    private static object ParseOperand(string text, Dictionary<string, Label> labels)
+    {
+        string token = text.Trim();
+
+        if (Registers.TryGetValue(token, out object? register))
+            return register;
+
+        if (labels.TryGetValue(token, out Label label))
+            return label;
+
+        if (TryParseImmediate(token, out object? value))
+            return new Immediate(value!);
+
+        throw new NotSupportedException(
+            $"Unrecognized operand '{token}' (expected a register, an immediate, or a label).");
+    }
+
+    // Parses an immediate as the narrowest of long/ulong that holds it, boxed. Storing the
+    // widest representation lets TryChangeType later narrow it to whatever integer parameter
+    // the chosen overload expects — and reject (rather than crash on) values that do not fit.
+    private static bool TryParseImmediate(string token, out object? value)
+    {
+        value = null;
+        bool negative = token.StartsWith('-');
+        string body = negative ? token[1..] : token;
+
+        if (body.StartsWith("0x", StringComparison.OrdinalIgnoreCase))
+        {
+            if (!ulong.TryParse(body[2..], NumberStyles.HexNumber, CultureInfo.InvariantCulture, out ulong hex))
+                return false;
+
+            value = negative ? -(long)hex : hex;
+            return true;
+        }
+
+        if (negative)
+        {
+            if (!long.TryParse(token, NumberStyles.Integer, CultureInfo.InvariantCulture, out long signed))
+                return false;
+
+            value = signed;
+            return true;
+        }
+
+        // Non-negative decimal: prefer long, fall back to ulong for values above long.MaxValue.
+        if (long.TryParse(body, NumberStyles.Integer, CultureInfo.InvariantCulture, out long asLong))
+            value = asLong;
+        else if (ulong.TryParse(body, NumberStyles.Integer, CultureInfo.InvariantCulture, out ulong asULong))
+            value = asULong;
+        else
+            return false;
+
+        return true;
+    }
+
+    private static bool TryChangeType(object value, Type targetType, out object? result)
+    {
+        try
+        {
+            result = Convert.ChangeType(value, targetType, CultureInfo.InvariantCulture);
+            return true;
+        }
+        catch (Exception ex) when (ex is OverflowException or InvalidCastException or FormatException)
+        {
+            result = null;
+            return false;
+        }
+    }
+
+    private static bool IsNumeric(Type type) => Type.GetTypeCode(type) is
+        TypeCode.SByte or TypeCode.Byte or TypeCode.Int16 or TypeCode.UInt16 or
+        TypeCode.Int32 or TypeCode.UInt32 or TypeCode.Int64 or TypeCode.UInt64;
+
+    private static List<(string Mnemonic, string[] Operands)> Tokenize(string text, out List<string> labelNames)
+    {
+        var result = new List<(string, string[])>();
+        labelNames = new List<string>();
+
+        foreach (string rawLine in text.Split('\n'))
+        {
+            string line = rawLine;
+
+            int comment = line.IndexOf(';');
+            if (comment >= 0)
+                line = line[..comment];
+
+            line = line.Trim();
+            if (line.Length == 0)
+                continue;
+
+            // A "name:" prefix is a label definition; keep any instruction that follows it
+            // on the same line as a separate entry.
+            int colon = line.IndexOf(':');
+            if (colon >= 0)
+            {
+                string labelName = line[..colon].Trim();
+                labelNames.Add(labelName);
+                result.Add((labelName + ":", Array.Empty<string>()));
+
+                line = line[(colon + 1)..].Trim();
+                if (line.Length == 0)
+                    continue;
+            }
+
+            int space = line.IndexOfAny([' ', '\t']);
+            if (space < 0)
+            {
+                result.Add((line, Array.Empty<string>()));
+                continue;
+            }
+
+            string mnemonic = line[..space];
+            string[] operands = line[(space + 1)..]
+                .Split(',', StringSplitOptions.RemoveEmptyEntries | StringSplitOptions.TrimEntries);
+            result.Add((mnemonic, operands));
+        }
+
+        return result;
+    }
+
+    private static Dictionary<string, object> BuildRegisterMap()
+    {
+        var map = new Dictionary<string, object>(StringComparer.OrdinalIgnoreCase);
+        foreach (FieldInfo field in typeof(AssemblerRegisters).GetFields(BindingFlags.Public | BindingFlags.Static))
+        {
+            object? value = field.GetValue(null);
+            if (value is not null)
+                map[field.Name] = value;
+        }
+
+        return map;
+    }
+
+    // A parsed immediate (boxed long or ulong), distinguished from register/label operands
+    // so binding can narrow it to whichever integer parameter type the chosen overload
+    // expects — or reject it when it does not fit.
+    private readonly record struct Immediate(object Value);
+
+    private sealed class ByteListCodeWriter : CodeWriter
+    {
+        public List<byte> Bytes { get; } = new();
+
+        public override void WriteByte(byte value) => Bytes.Add(value);
+    }
+}

WhiteMagic/Discovery/PeHeaderParser.cs

@@ -1,5 +1,6 @@
 using System.ComponentModel;
 using System.Runtime.InteropServices;
+using System.Text;
 using WhiteMagic.Native;
 
 namespace WhiteMagic.Discovery;
@@ -103,6 +104,150 @@ public sealed class PeHeaderParser
         }
     }
 
+    /// <summary>
+    /// Resolves an exported function's absolute address by name, following export
+    /// forwarders (e.g. <c>kernel32!HeapAlloc</c> → <c>NTDLL.RtlAllocateHeap</c>) into
+    /// other modules loaded in the same target process.
+    /// </summary>
+    /// <param name="functionName">The exported symbol name (case-sensitive, as stored
+    /// in the export name table).</param>
+    /// <returns>The absolute address of the export in the target process.</returns>
+    /// <remarks>
+    /// Forwarders are resolved by locating the target module in the process's loaded-module
+    /// list. API-set forwarders (virtual <c>api-ms-win-*</c> / <c>ext-ms-*</c> names) are NOT
+    /// supported: those are not real loaded modules, so resolution through the module list is
+    /// impossible without parsing the API-set schema — such a forwarder throws
+    /// <see cref="NotSupportedException"/>. On modern Windows many system-DLL exports forward
+    /// through API sets; resolve those via the OS loader (<c>GetProcAddress</c>) instead.
+    /// Ordinal forwarders (<c>Module.#N</c>) are likewise unsupported.
+    /// </remarks>
+    /// <exception cref="InvalidOperationException">The export is not present.</exception>
+    /// <exception cref="NotSupportedException">The export forwards to an ordinal or to a
+    /// module (such as an API set) that is not resolvable from the target's module list.</exception>
+    /// <exception cref="InvalidDataException">The PE export data is malformed.</exception>
+    public IntPtr GetExportAddress(string functionName)
+    {
+        ArgumentException.ThrowIfNullOrEmpty(functionName);
+        return ResolveExport(functionName, 0);
+    }
+
+    // Maximum forwarder hops before giving up, to bound pathological chains.
+    private const int MaxForwarderDepth = 16;
+
+    private IntPtr ResolveExport(string functionName, int depth)
+    {
+        if (depth > MaxForwarderDepth)
+            throw new InvalidDataException($"Export forwarder chain for '{functionName}' is too deep.");
+
+        var (optionalHeader, _) = ParseOptionalHeader();
+        if (optionalHeader is null || optionalHeader.Length < 2)
+            throw new InvalidDataException("Optional header unavailable.");
+
+        // 0x10b = PE32 (32-bit), 0x20b = PE32+ (64-bit). Data directories start at a
+        // different offset in each: 96 for PE32, 112 for PE32+. The export table is
+        // directory index 0, so its 8-byte entry sits at that offset.
+        ushort magic = BitConverter.ToUInt16(optionalHeader, 0);
+        bool pe32Plus = magic == 0x20b;
+        int exportDirOffset = pe32Plus ? 112 : 96;
+        if (optionalHeader.Length < exportDirOffset + 8)
+            throw new InvalidDataException("Optional header does not contain the export data directory.");
+
+        uint exportRva = BitConverter.ToUInt32(optionalHeader, exportDirOffset);
+        uint exportSize = BitConverter.ToUInt32(optionalHeader, exportDirOffset + 4);
+        if (exportRva == 0 || exportSize == 0)
+            throw new InvalidOperationException("Module has no export table.");
+
+        // IMAGE_EXPORT_DIRECTORY is 40 bytes.
+        byte[] dir = _memory.ReadBytes(_baseAddress + (nint)exportRva, 40);
+        if (dir.Length < 40)
+            throw new InvalidDataException("Failed to read the export directory.");
+
+        uint numberOfFunctions = BitConverter.ToUInt32(dir, 20);
+        uint numberOfNames = BitConverter.ToUInt32(dir, 24);
+        uint addressOfFunctions = BitConverter.ToUInt32(dir, 28);
+        uint addressOfNames = BitConverter.ToUInt32(dir, 32);
+        uint addressOfNameOrdinals = BitConverter.ToUInt32(dir, 36);
+
+        // Guard against corrupt counts before allocating arrays sized from them.
+        if (numberOfNames > 0x10000 || numberOfFunctions > 0x10000)
+            throw new InvalidDataException("Export table entry count is out of range.");
+
+        if (numberOfNames == 0)
+            throw new InvalidOperationException($"Export '{functionName}' not found (module exports no names).");
+
+        byte[] nameRvas = _memory.ReadBytes(_baseAddress + (nint)addressOfNames, checked((int)(numberOfNames * 4)));
+        byte[] nameOrdinals = _memory.ReadBytes(_baseAddress + (nint)addressOfNameOrdinals, checked((int)(numberOfNames * 2)));
+        if (nameRvas.Length < numberOfNames * 4 || nameOrdinals.Length < numberOfNames * 2)
+            throw new InvalidDataException("Failed to read the export name tables.");
+
+        int nameIndex = -1;
+        for (int i = 0; i < numberOfNames; i++)
+        {
+            uint nameRva = BitConverter.ToUInt32(nameRvas, i * 4);
+            string name = _memory.ReadString(_baseAddress + (nint)nameRva, Encoding.ASCII, 512);
+            if (string.Equals(name, functionName, StringComparison.Ordinal))
+            {
+                nameIndex = i;
+                break;
+            }
+        }
+
+        if (nameIndex < 0)
+            throw new InvalidOperationException($"Export '{functionName}' not found in module.");
+
+        ushort ordinal = BitConverter.ToUInt16(nameOrdinals, nameIndex * 2);
+        if (ordinal >= numberOfFunctions)
+            throw new InvalidDataException("Export name ordinal is out of range.");
+
+        byte[] funcRvaBytes = _memory.ReadBytes(
+            _baseAddress + (nint)(addressOfFunctions + (uint)ordinal * 4u), 4);
+        if (funcRvaBytes.Length < 4)
+            throw new InvalidDataException("Failed to read the export address table entry.");
+
+        uint funcRva = BitConverter.ToUInt32(funcRvaBytes, 0);
+        if (funcRva == 0)
+            throw new InvalidOperationException($"Export '{functionName}' has no address.");
+
+        // A function RVA that lands inside the export directory region is not code but a
+        // null-terminated "Module.Function" forwarder string.
+        if (funcRva >= exportRva && funcRva < exportRva + exportSize)
+        {
+            string forwarder = _memory.ReadString(_baseAddress + (nint)funcRva, Encoding.ASCII, 512);
+            return ResolveForwarder(forwarder, depth);
+        }
+
+        return _baseAddress + (nint)funcRva;
+    }
+
+    private IntPtr ResolveForwarder(string forwarder, int depth)
+    {
+        // A forwarder is "Module.Function"; the module name carries no extension, so the
+        // FIRST dot is the boundary. Splitting on the last dot would misparse export names
+        // that themselves contain a dot (e.g. some C++/managed exports).
+        int dot = forwarder.IndexOf('.');
+        if (dot <= 0 || dot >= forwarder.Length - 1)
+            throw new InvalidDataException($"Malformed export forwarder string '{forwarder}'.");
+
+        string moduleName = forwarder[..dot];
+        string target = forwarder[(dot + 1)..];
+
+        if (target.StartsWith('#'))
+        {
+            throw new NotSupportedException(
+                $"Ordinal export forwarders are not supported (forwarder '{forwarder}').");
+        }
+
+        IntPtr targetBase = RemoteModule.ResolveBase(_memory.ProcessId, moduleName);
+        if (targetBase == IntPtr.Zero)
+        {
+            throw new NotSupportedException(
+                $"Export forwarder target module '{moduleName}' is not loaded in the target " +
+                $"process, or is an unresolvable API set (forwarder '{forwarder}').");
+        }
+
+        return new PeHeaderParser(_memory, targetBase).ResolveExport(target, depth + 1);
+    }
+
     /// <summary>
     /// Parses the DOS header, PE signature, and optional header.
     /// </summary>

WhiteMagic/Hooking/Detour.cs

@@ -16,6 +16,7 @@ namespace WhiteMagic.Hooking;
 public sealed class Detour : IDisposable
 {
     private readonly MemoryBase _memory;
+    private readonly PrologueLengthResolver _prologueLength;
 
     /// <summary>The unique name of this detour.</summary>
     public string Name { get; }
@@ -43,14 +44,21 @@ public sealed class Detour : IDisposable
     /// <summary><see langword="true"/> while the detour bytes are live at <see cref="Target"/>.</summary>
     public bool IsApplied { get; private set; }
 
-    internal Detour(MemoryBase memory, string name, IntPtr target, Delegate hook)
+    internal Detour(
+        MemoryBase memory,
+        string name,
+        IntPtr target,
+        Delegate hook,
+        PrologueLengthResolver prologueLength)
     {
         ArgumentNullException.ThrowIfNull(hook);
+        ArgumentNullException.ThrowIfNull(prologueLength);
 
         _memory = memory;
         Name = name;
         Target = target;
         Hook = hook;
+        _prologueLength = prologueLength;
     }
 
     /// <summary>
@@ -83,7 +91,7 @@ public sealed class Detour : IDisposable
                 "Could not read enough bytes from the target function to install a detour.");
         }
 
-        int preserveLength = PrologueDecoder.GetWholeInstructionLength(prologue, detourLength, _memory.Is64Bit);
+        int preserveLength = _prologueLength(prologue, detourLength, _memory.Is64Bit);
         OverwrittenBytes = new byte[preserveLength];
         Buffer.BlockCopy(prologue, 0, OverwrittenBytes, 0, preserveLength);
 

WhiteMagic/Hooking/DetourManager.cs

@@ -19,6 +19,15 @@ public sealed class DetourManager
         _memory = memory;
     }
 
+    /// <summary>
+    /// Resolves how many whole prologue-instruction bytes a splice must preserve. Defaults
+    /// to the built-in <see cref="PrologueDecoder"/>, which covers only the common prologue
+    /// shapes and rejects anything else. Assign <c>new IcedAssembler().GetPrologueLength</c>
+    /// to validate arbitrary prologues via the optional Iced disassembler.
+    /// </summary>
+    public PrologueLengthResolver PrologueLengthResolver { get; set; } =
+        PrologueDecoder.GetWholeInstructionLength;
+
     /// <summary>
     /// Creates a new detour and registers it with the manager.
     /// The <paramref name="hook"/> delegate's type must match the native signature of
@@ -26,7 +35,7 @@ public sealed class DetourManager
     /// </summary>
     public Detour Create(string name, IntPtr target, Delegate hook)
     {
-        var detour = new Detour(_memory, name, target, hook);
+        var detour = new Detour(_memory, name, target, hook, PrologueLengthResolver);
         _detours[name] = detour;
         return detour;
     }

WhiteMagic/Hooking/PrologueDecoder.cs

@@ -2,6 +2,14 @@ using System;
 
 namespace WhiteMagic.Hooking;
 
+/// <summary>
+/// Resolves how many whole prologue-instruction bytes must be preserved to splice
+/// <paramref name="requiredBytes"/> bytes at a target. The built-in
+/// <see cref="PrologueDecoder.GetWholeInstructionLength"/> satisfies this delegate, as
+/// does <c>IcedAssembler.GetPrologueLength</c> for full instruction coverage.
+/// </summary>
+public delegate int PrologueLengthResolver(byte[] prologue, int requiredBytes, bool is64Bit);
+
 /// <summary>
 /// Minimal instruction-length decoder for common x86/x64 prologue shapes.
 /// The set is intentionally small: any opcode outside the covered set is rejected

WhiteMagic/Magic.cs

@@ -54,6 +54,10 @@ public sealed class Magic : IDisposable
     /// <summary>Returns a <see cref="RemotePointer"/> at <paramref name="address"/>.</summary>
     public RemotePointer this[IntPtr address] => new RemotePointer(Memory, address);
 
+    /// <summary>Returns the loaded <see cref="RemoteModule"/> named <paramref name="moduleName"/>
+    /// (e.g. <c>magic["user32"]["MessageBoxA"]</c>).</summary>
+    public RemoteModule this[string moduleName] => new RemoteModule(this, moduleName);
+
     /// <inheritdoc />
     public void Dispose()
     {

WhiteMagic/RemoteFunction.cs

@@ -0,0 +1,73 @@
+using System.Threading.Tasks;
+using WhiteMagic.Assembly;
+using WhiteMagic.Execution;
+
+namespace WhiteMagic;
+
+/// <summary>
+/// An exported function resolved in the target process, obtained via
+/// <c>magic["module"]["function"]</c>. Executes through one of the session's execution
+/// strategies.
+/// </summary>
+/// <remarks>
+/// The default <see cref="Execute{T}"/> path uses the always-available
+/// <see cref="RemoteThreadExecutor"/> (<c>CreateRemoteThread</c>), which is safe for
+/// thread-agnostic exports. For a call that touches single-threaded target state, obtain
+/// the <see cref="Address"/> and route it through a <see cref="MainThreadPump"/>, or use
+/// <see cref="CreateDelegate{TDelegate}"/> when running in-process.
+/// </remarks>
+public sealed class RemoteFunction
+{
+    private readonly Magic _magic;
+
+    /// <summary>The export name this function was resolved from.</summary>
+    public string Name { get; }
+
+    /// <summary>The absolute address of the function in the target process.</summary>
+    public IntPtr Address { get; }
+
+    internal RemoteFunction(Magic magic, string name, IntPtr address)
+    {
+        _magic = magic;
+        Name = name;
+        Address = address;
+    }
+
+    /// <summary>
+    /// Calls the function via a remote thread and returns its result cast to
+    /// <typeparamref name="T"/>.
+    /// </summary>
+    /// <param name="convention">The calling convention (ignored on x64 targets).</param>
+    /// <param name="args">Arguments to pass; primitives, pointers, enums, strings and
+    /// structs are supported.</param>
+    public T Execute<T>(CallConvention convention, params object?[] args)
+    {
+        return _magic.RemoteThread.Execute<T>(Address, convention, args);
+    }
+
+    /// <summary>Asynchronous variant of <see cref="Execute{T}"/>.</summary>
+    public Task<T> ExecuteAsync<T>(CallConvention convention, params object?[] args)
+    {
+        return _magic.RemoteThread.ExecuteAsync<T>(Address, convention, args);
+    }
+
+    /// <summary>
+    /// Creates a managed delegate bound to this function for the in-process scenario.
+    /// </summary>
+    /// <exception cref="InvalidOperationException">The session is not in-process. The
+    /// resolved <see cref="Address"/> lives in the target process; a delegate to it would
+    /// access-violate when invoked from the host, so this is rejected for external sessions.
+    /// Use <see cref="Execute{T}"/> (remote thread) for external targets.</exception>
+    public TDelegate CreateDelegate<TDelegate>() where TDelegate : Delegate
+    {
+        if (_magic.Memory is not InProcessReader)
+        {
+            throw new InvalidOperationException(
+                "CreateDelegate is only valid for an in-process session (Magic.OpenInProcess). " +
+                "The function address is not mapped into the host process for an external target; " +
+                "use Execute<T> to call it via a remote thread.");
+        }
+
+        return new InProcessInvoker(_magic.Memory).CreateFunction<TDelegate>(Address);
+    }
+}

WhiteMagic/RemoteModule.cs

@@ -0,0 +1,101 @@
+using WhiteMagic.Discovery;
+using Process = System.Diagnostics.Process;
+using ProcessModule = System.Diagnostics.ProcessModule;
+
+namespace WhiteMagic;
+
+/// <summary>
+/// A module (loaded DLL/EXE image) in the target process, obtained by indexing the
+/// facade with a module name (e.g. <c>magic["user32"]</c>). Exposes the module's base
+/// address and resolves exported functions by name.
+/// </summary>
+public sealed class RemoteModule
+{
+    private readonly Magic _magic;
+
+    /// <summary>The module's file name as reported by the OS (e.g. <c>user32.dll</c>).</summary>
+    public string Name { get; }
+
+    /// <summary>The module's load address in the target process.</summary>
+    public IntPtr BaseAddress { get; }
+
+    internal RemoteModule(Magic magic, string moduleName)
+    {
+        ArgumentNullException.ThrowIfNull(magic);
+        ArgumentException.ThrowIfNullOrEmpty(moduleName);
+
+        _magic = magic;
+
+        (string name, IntPtr baseAddress) = FindModule(magic.Memory.ProcessId, moduleName);
+        Name = name;
+        BaseAddress = baseAddress;
+    }
+
+    /// <summary>
+    /// Resolves an exported function by name and returns a <see cref="RemoteFunction"/>
+    /// bound to its address. Export forwarders are followed.
+    /// </summary>
+    public RemoteFunction this[string functionName]
+    {
+        get
+        {
+            IntPtr address = GetExportAddress(functionName);
+            return new RemoteFunction(_magic, functionName, address);
+        }
+    }
+
+    /// <summary>Resolves the absolute address of an exported function by name.</summary>
+    public IntPtr GetExportAddress(string functionName)
+    {
+        ArgumentException.ThrowIfNullOrEmpty(functionName);
+        var parser = new PeHeaderParser(_magic.Memory, BaseAddress);
+        return parser.GetExportAddress(functionName);
+    }
+
+    private static (string Name, IntPtr BaseAddress) FindModule(int processId, string moduleName)
+    {
+        using Process process = Process.GetProcessById(processId);
+        foreach (ProcessModule module in process.Modules)
+        {
+            if (NameMatches(module.ModuleName, moduleName))
+                return (module.ModuleName, module.BaseAddress);
+        }
+
+        throw new DllNotFoundException(
+            $"Module '{moduleName}' is not loaded in process {processId}.");
+    }
+
+    /// <summary>
+    /// Resolves a module's base address by name within a target process, returning
+    /// <see cref="IntPtr.Zero"/> if it is not loaded. Used by export-forwarder resolution.
+    /// </summary>
+    internal static IntPtr ResolveBase(int processId, string moduleName)
+    {
+        using Process process = Process.GetProcessById(processId);
+        foreach (ProcessModule module in process.Modules)
+        {
+            if (NameMatches(module.ModuleName, moduleName))
+                return module.BaseAddress;
+        }
+
+        return IntPtr.Zero;
+    }
+
+    /// <summary>
+    /// Matches a loaded module's file name against a requested name, tolerating a missing
+    /// or present <c>.dll</c> extension and ignoring case (e.g. <c>KERNEL32</c> matches
+    /// <c>kernel32.dll</c>).
+    /// </summary>
+    private static bool NameMatches(string actual, string requested)
+    {
+        if (string.Equals(actual, requested, StringComparison.OrdinalIgnoreCase))
+            return true;
+
+        string actualNoExt = Path.GetFileNameWithoutExtension(actual);
+        string requestedNoExt = requested.EndsWith(".dll", StringComparison.OrdinalIgnoreCase)
+            ? requested[..^4]
+            : requested;
+
+        return string.Equals(actualNoExt, requestedNoExt, StringComparison.OrdinalIgnoreCase);
+    }
+}

WhiteMagic/WhiteMagic.csproj

@@ -13,4 +13,13 @@
     <InternalsVisibleTo Include="WhiteMagicTest" />
   </ItemGroup>
 
+  <!--
+    Optional Iced backend (task 8.x). Isolated behind IAssembler: the default
+    StubAssembler path never touches Iced, keeping the common configuration free of any
+    behavioral dependency on it. Only callers that construct IcedAssembler pull it in.
+  -->
+  <ItemGroup>
+    <PackageReference Include="Iced" Version="1.21.0" />
+  </ItemGroup>
+
 </Project>

WhiteMagicTest/Assembly/IcedAssemblerTests.cs

@@ -0,0 +1,159 @@
+using System.Linq;
+using Iced.Intel;
+using WhiteMagic;
+using WhiteMagic.Assembly;
+using WhiteMagic.Hooking;
+
+namespace WhiteMagicTest.Assembly;
+
+/// <summary>
+/// Tests for the optional <see cref="IcedAssembler"/> backend (tasks 8.1–8.3): arbitrary
+/// text assembly, origin-relative encoding, and full prologue instruction decoding.
+/// </summary>
+public class IcedAssemblerTests
+{
+    private static Instruction[] Disassemble(byte[] code, int bitness, ulong origin)
+    {
+        var decoder = Decoder.Create(bitness, new ByteArrayCodeReader(code));
+        decoder.IP = origin;
+
+        var result = new List<Instruction>();
+        ulong end = origin + (ulong)code.Length;
+        while (decoder.IP < end)
+            result.Add(decoder.Decode());
+
+        return result.ToArray();
+    }
+
+    [Fact]
+    public void Assemble_emits_single_instruction()
+    {
+        var assembler = new IcedAssembler(64);
+        byte[] code = assembler.Assemble("ret");
+        Assert.Equal(new byte[] { 0xC3 }, code);
+    }
+
+    [Fact]
+    public void Assemble_emits_multiple_instructions_with_operands()
+    {
+        var assembler = new IcedAssembler(32);
+
+        // The scenario from the managed-assembler spec.
+        byte[] code = assembler.Assemble("push 0\nadd esp, 4\nret");
+        Assert.NotEmpty(code);
+
+        Instruction[] instructions = Disassemble(code, 32, 0);
+        Assert.Equal(3, instructions.Length);
+        Assert.Equal(Mnemonic.Push, instructions[0].Mnemonic);
+        Assert.Equal(Mnemonic.Add, instructions[1].Mnemonic);
+        Assert.Equal(Register.ESP, instructions[1].Op0Register);
+        Assert.Equal(4UL, instructions[1].GetImmediate(1));
+        Assert.Equal(Mnemonic.Ret, instructions[2].Mnemonic);
+    }
+
+    [Fact]
+    public void Assemble_supports_comments_and_blank_lines()
+    {
+        var assembler = new IcedAssembler(64);
+        byte[] code = assembler.Assemble("  ; prologue\n\nnop ; a comment\nret\n");
+
+        Instruction[] instructions = Disassemble(code, 64, 0);
+        Assert.Equal(2, instructions.Length);
+        Assert.Equal(Mnemonic.Nop, instructions[0].Mnemonic);
+        Assert.Equal(Mnemonic.Ret, instructions[1].Mnemonic);
+    }
+
+    [Fact]
+    public void Assemble_encodes_label_branch_relative_to_origin()
+    {
+        var assembler = new IcedAssembler(64);
+        const ulong origin = 0x1_4000_1000UL;
+
+        // jmp forward over a nop to a label; the near-branch target must be resolved
+        // against the supplied origin, not zero.
+        byte[] code = assembler.Assemble("jmp done\nnop\ndone:\nret", origin);
+
+        Instruction[] instructions = Disassemble(code, 64, origin);
+        Instruction jmp = instructions[0];
+        Assert.Equal(Mnemonic.Jmp, jmp.Mnemonic);
+
+        // Target = origin + len(jmp) + len(nop): the address of the 'done: ret'.
+        ulong expected = origin + (ulong)jmp.Length + 1;
+        Assert.Equal(expected, jmp.NearBranchTarget);
+    }
+
+    [Theory]
+    [InlineData("mov eax, 4294967295")]   // 0xFFFFFFFF — needs the uint overload, not int
+    [InlineData("mov eax, 0xFFFFFFFF")]    // same value, hex form
+    [InlineData("mov rax, 18446744073709551615")] // ulong.MaxValue — decimal above long.MaxValue
+    public void Assemble_binds_wide_unsigned_immediates(string source)
+    {
+        var assembler = new IcedAssembler(64);
+
+        byte[] code = assembler.Assemble(source);
+        Assert.NotEmpty(code);
+
+        Instruction[] instructions = Disassemble(code, 64, 0);
+        Assert.Single(instructions);
+        Assert.Equal(Mnemonic.Mov, instructions[0].Mnemonic);
+    }
+
+    [Fact]
+    public void Assemble_rejects_immediate_that_fits_no_overload_without_crashing()
+    {
+        var assembler = new IcedAssembler(64);
+
+        // -2147483649 is below int.MinValue and eax has no wider signed overload; must be a
+        // clean NotSupportedException, not an OverflowException escaping from ChangeType.
+        Assert.Throws<NotSupportedException>(() => assembler.Assemble("mov eax, -2147483649"));
+    }
+
+    [Fact]
+    public void Assemble_throws_on_unsupported_operand()
+    {
+        var assembler = new IcedAssembler(64);
+        Assert.Throws<NotSupportedException>(() => assembler.Assemble("mov rax, [rbx]"));
+    }
+
+    [Fact]
+    public void GetPrologueLength_decodes_prologue_the_builtin_decoder_rejects()
+    {
+        // 48 8B C1 = mov rax, rcx — a register-to-register mov the built-in PrologueDecoder
+        // does not cover (it only recognizes the 8B FF / 8B EC forms).
+        // Followed by push rbp; mov rbp,rsp; sub rsp,0x20; mov rax,rcx to exceed 14 bytes.
+        byte[] prologue =
+        [
+            0x48, 0x8B, 0xC1,             // mov rax, rcx      (3)
+            0x55,                         // push rbp          (1)
+            0x48, 0x8B, 0xEC,             // mov rbp, rsp      (3)
+            0x48, 0x83, 0xEC, 0x20,       // sub rsp, 0x20     (4)
+            0x48, 0x8B, 0xC1              // mov rax, rcx      (3)  -> total 14
+        ];
+
+        // The built-in decoder refuses the very first instruction.
+        Assert.Throws<InvalidOperationException>(() =>
+            PrologueDecoder.GetWholeInstructionLength(prologue, 14, is64Bit: true));
+
+        // The Iced backend decodes it and returns the whole-instruction length covering
+        // at least the 14 bytes a detour needs.
+        var iced = new IcedAssembler();
+        int length = iced.GetPrologueLength(prologue, 14, is64Bit: true);
+        Assert.Equal(14, length);
+    }
+
+    [Fact]
+    public void DetourManager_prologue_resolver_defaults_to_builtin_and_is_replaceable()
+    {
+        using var reader = new InProcessReader();
+        var manager = new DetourManager(reader);
+
+        // Default resolver is the built-in decoder.
+        Assert.Throws<InvalidOperationException>(() =>
+            manager.PrologueLengthResolver(new byte[] { 0x48, 0x8B, 0xC1, 0x90, 0x90 }, 4, true));
+
+        // Swapping in the Iced resolver validates the same bytes.
+        manager.PrologueLengthResolver = new IcedAssembler().GetPrologueLength;
+        int length = manager.PrologueLengthResolver(new byte[] { 0x48, 0x8B, 0xC1, 0x90, 0x90 }, 4, true);
+        Assert.True(length >= 4);
+    }
+}

WhiteMagicTest/Execution/RemoteThreadExecutorTests.cs

@@ -35,6 +35,33 @@ public sealed class RemoteThreadExecutorTests
         0xC3
     ];
 
+    // Five-arg callee that also executes an alignment-sensitive SSE instruction, proving
+    // the stub delivers a 16-byte-aligned stack the CPU actually accepts (movaps #GPs on a
+    // misaligned address) alongside correct register+stack argument placement.
+    // sub rsp, 24           ; entry rsp ≡ 8 (mod 16) -> rsp ≡ 0 (16-aligned), giving a
+    //                       ; 16-byte aligned scratch at [rsp..rsp+16) below the saved
+    //                       ; return address ([rsp+24]) so the store leaves it intact
+    // movaps [rsp], xmm0    ; aligned 16-byte store — faults unless rsp is 16-aligned
+    // add rsp, 24           ; restore
+    // mov eax, ecx
+    // add eax, edx
+    // add eax, r8d
+    // add eax, r9d
+    // add eax, [rsp+0x28]   ; 5th arg above the shadow space
+    // ret
+    private static readonly byte[] SseAlignedSumPayload =
+    [
+        0x48, 0x83, 0xEC, 0x18,
+        0x0F, 0x29, 0x04, 0x24,
+        0x48, 0x83, 0xC4, 0x18,
+        0x89, 0xC8,
+        0x01, 0xD0,
+        0x44, 0x01, 0xC0,
+        0x44, 0x01, 0xC8,
+        0x03, 0x84, 0x24, 0x28, 0x00, 0x00, 0x00,
+        0xC3
+    ];
+
     // xor eax, eax
     // cmp byte ptr [rcx+rax], 0
     // je done
@@ -110,6 +137,21 @@ public sealed class RemoteThreadExecutorTests
         Assert.Equal(0, misalign);
     }
 
+    [Fact]
+    public void Execute_runs_sse_callee_with_five_args()
+    {
+        if (!Environment.Is64BitProcess)
+        {
+            return;
+        }
+
+        // Correct result (150) requires BOTH the 5th arg reaching [rsp+0x28] AND the
+        // aligned movaps not faulting. A broken frame size/alignment either mis-sums or
+        // #GPs in the callee.
+        int result = RunPayload(SseAlignedSumPayload, CallConvention.Cdecl, 10, 20, 30, 40, 50);
+        Assert.Equal(150, result);
+    }
+
     [Fact]
     public void Execute_marshals_string_as_utf8_pointer()
     {

WhiteMagicTest/ModuleFunctionTests.cs

@@ -0,0 +1,121 @@
+using System;
+using System.Diagnostics;
+using WhiteMagic;
+using WhiteMagic.Assembly;
+using WhiteMagic.Native;
+using Xunit;
+
+namespace WhiteMagicTest;
+
+/// <summary>
+/// Tests for <see cref="RemoteModule"/> / <see cref="RemoteFunction"/> resolution and
+/// execution through the <see cref="Magic"/> facade (task 7.2).
+/// </summary>
+public class ModuleFunctionTests
+{
+    // Ensure a module is loaded in this process before resolving it.
+    private static IntPtr Load(string module)
+    {
+        IntPtr handle = NativeMethods.LoadLibrary(module);
+        Assert.NotEqual(IntPtr.Zero, handle);
+        return handle;
+    }
+
+    [Fact]
+    public void Module_indexer_resolves_base_address()
+    {
+        IntPtr handle = Load("kernel32.dll");
+
+        using var magic = Magic.OpenInProcess();
+        RemoteModule module = magic["kernel32"];
+
+        // The module handle returned by LoadLibrary is the module's base address.
+        Assert.Equal(handle, module.BaseAddress);
+        Assert.Equal("KERNEL32.DLL", module.Name, ignoreCase: true);
+    }
+
+    [Fact]
+    public void Function_indexer_resolves_direct_export()
+    {
+        IntPtr handle = Load("user32.dll");
+        IntPtr expected = NativeMethods.GetProcAddress(handle, "MessageBoxA");
+        Assert.NotEqual(IntPtr.Zero, expected);
+
+        using var magic = Magic.OpenInProcess();
+        RemoteFunction fn = magic["user32"]["MessageBoxA"];
+
+        Assert.Equal(expected, fn.Address);
+        Assert.Equal("MessageBoxA", fn.Name);
+    }
+
+    [Fact]
+    public void Function_indexer_follows_export_forwarder()
+    {
+        // kernel32!HeapAlloc is a classic forwarder to NTDLL.RtlAllocateHeap. Whatever the
+        // OS loader resolves it to, our parser must reach the same final address.
+        IntPtr handle = Load("kernel32.dll");
+        IntPtr expected = NativeMethods.GetProcAddress(handle, "HeapAlloc");
+        Assert.NotEqual(IntPtr.Zero, expected);
+
+        using var magic = Magic.OpenInProcess();
+        RemoteFunction fn = magic["kernel32"]["HeapAlloc"];
+
+        Assert.Equal(expected, fn.Address);
+    }
+
+    [Fact]
+    public void Module_indexer_throws_for_unloaded_module()
+    {
+        using var magic = Magic.OpenInProcess();
+        Assert.Throws<DllNotFoundException>(() => magic["definitely-not-loaded-xyz.dll"]);
+    }
+
+    [Fact]
+    public void Function_indexer_throws_for_unknown_export()
+    {
+        Load("kernel32.dll");
+
+        using var magic = Magic.OpenInProcess();
+        Assert.Throws<InvalidOperationException>(() => magic["kernel32"]["NoSuchExport_ZZZ"]);
+    }
+
+    private delegate uint GetCurrentProcessIdDelegate();
+
+    [Fact]
+    public void CreateDelegate_throws_for_external_session()
+    {
+        Load("kernel32.dll");
+
+        // External reader (even to self): the address is not treated as host-mapped, so a
+        // delegate to it is rejected rather than handed back to AV on invocation.
+        using var magic = Magic.Open(Process.GetCurrentProcess());
+        RemoteFunction fn = magic["kernel32"]["GetCurrentProcessId"];
+
+        Assert.Throws<InvalidOperationException>(() => fn.CreateDelegate<GetCurrentProcessIdDelegate>());
+    }
+
+    [Fact]
+    public void CreateDelegate_invokes_function_in_process()
+    {
+        Load("kernel32.dll");
+
+        using var magic = Magic.OpenInProcess();
+        var getPid = magic["kernel32"]["GetCurrentProcessId"].CreateDelegate<GetCurrentProcessIdDelegate>();
+
+        Assert.Equal((uint)Process.GetCurrentProcess().Id, getPid());
+    }
+
+    [Fact]
+    public void Resolved_function_executes_via_remote_thread()
+    {
+        Load("kernel32.dll");
+
+        using var magic = Magic.OpenInProcess();
+        RemoteFunction getPid = magic["kernel32"]["GetCurrentProcessId"];
+
+        // GetCurrentProcessId takes no args and is thread-agnostic; a remote thread in our
+        // own process must report our PID.
+        uint pid = getPid.Execute<uint>(CallConvention.Stdcall);
+        Assert.Equal((uint)Process.GetCurrentProcess().Id, pid);
+    }
+}

docs/memory-library-comparison.md

@@ -74,7 +74,7 @@ WhiteMagic (facade — BM-old ergonomics)
 ├─ Core:        SafeHandle, native P/Invoke, x64             [BM current]
 ├─ MemoryBase (abstract Read/Write + MarshalCache)           [GreyMagic]
 │   ├─ ExternalReader (RPM/WPM)
-│   └─ InProcessReader (direct deref, injected)
+│   └─ InProcessReader (RPM/WPM on self-handle, injected)
 ├─ Discovery:   PatternScanner(+cache), PeHeaderParser       [BM current + GreyMagic]
 ├─ Allocation:  AllocatedMemory (named chunks)               [GreyMagic]
 ├─ Assembler:   IAssembler → { HandStubs | Iced }            [BM current; Iced replaces FASM]
@@ -91,3 +91,16 @@ WhiteMagic (facade — BM-old ergonomics)
 ```
 
 **Net result**: BM's modern, FASM-free, x64 core + GreyMagic's dual-mode / detour / patch / marshal-cache engine + MemorySharp's high-level ergonomics — with a three-tier execution model whose *default* for state-sensitive calls is the crash-safe main-thread pump, while `CreateRemoteThread` stays available for the payloads it is genuinely safe for.
+
+### Deviations discovered during implementation
+
+The design held, but building it surfaced corrections worth recording (each is detailed against its task in `openspec/changes/whitemagic-foundation/tasks.md`):
+
+- **`InProcessReader` reads via RPM/WPM on a self-handle, not `unsafe` direct deref** — .NET cannot catch `AccessViolationException`, so a bad direct deref kills the host with no soft-failure path. The in-process speed win moves to the delegate-call and detour paths, not the reader (design decision D1, revised mid-Phase 2).
+- **`MarshalCache<T>` splits `Size` (managed, blittable) from `MarshalSize` (`Marshal.SizeOf`, marshal path)** — a single size mis-sized structs whose unmanaged width differs (a `bool` field is managed-1 / unmanaged-4; inline `ByValTStr`/`ByValArray` under-sized the marshal buffer and corrupted the heap on write). `MemoryBase` picks per `TypeRequiresMarshal` at every IO site.
+- **x64 call stub is fully MS-x64-ABI compliant** — 32-byte shadow space, 16-byte alignment at the inner `call`, full `imm64` register loads (no >4 GiB pointer truncation), stack args above the shadow window. Proven at runtime by a live SSE callee whose aligned `movaps` faults on any misalignment (task 3.8), not just by byte-level encoding tests.
+- **`RemoteModule`/`RemoteFunction` follow PE export forwarders** — `kernel32!HeapAlloc` → `NTDLL.RtlAllocateHeap` and similar resolve into the real target module; ordinal and API-set forwarders throw `NotSupportedException` rather than returning a wrong address (task 7.2).
+- **Detour prologue safety is tiered** — the default `StubAssembler` length-decoder covers only the common x86/x64 prologue shapes and refuses any opcode outside that set (zero dependency); the optional `IcedAssembler.GetPrologueLength` decodes arbitrary prologues and is plugged in via `DetourManager.PrologueLengthResolver` when full validation is wanted (tasks 4.6, 8.3).
+- **Iced has no text parser** — the design assumed arbitrary text assembly could be delegated to Iced, but Iced ships only a *fluent* code assembler and a decoder. `IcedAssembler.Assemble` bridges Intel-syntax text onto the fluent API by reflection (registers, immediates, labels; memory operands unsupported), rather than depending on a parser that does not exist (task 8.2).
+- **Injection bitness corrections** — the thread-hijack injector enforces matching host/target bitness, so the 32-bit path always runs from a 32-bit caller and uses native `GetThreadContext`/`SetThreadContext`; the WOW64 context APIs (for 64-bit callers inspecting WOW64 targets) never apply here and were removed. `ExternalReader` validates `QueryInformation`/`QueryLimitedInformation` access and surfaces `IsWow64Process` failures instead of silently assuming host bitness.
+- **Bounds and protection hardening** — `AllocatedMemory` range-checks typed IO against region size; `Patch` mirrors the detour's `VirtualProtectEx` dance; `MainThreadPump` guards the completion race on an already-completed `TaskCompletionSource`.

openspec/changes/whitemagic-foundation/tasks.md

@@ -28,7 +28,7 @@
 - [x] 3.5 Add tests for stdcall (no caller cleanup), thiscall (ecx = this), fastcall (ecx/edx) x86 stubs
 - [x] 3.6 Implement x86 stdcall/thiscall/fastcall stubs to pass 3.5
 - [x] 3.7 Add tests for x64 stub argument-register placement and call
-- [x] 3.8 Implement x64 stub to pass 3.7. **Deviation (review):** `BuildCallStub` takes `nuint[]` (was `uint[]`). x64 stub is Microsoft-x64-ABI compliant: allocates 32-byte shadow space, keeps 16-byte stack alignment at the inner `call` (frame `K ≡ 8 (mod 16)`, `K ≥ 0x20 + 8·stackArgs`), loads RCX/RDX/R8/R9 with full 64-bit `imm64` (no >4 GiB pointer truncation), and writes stack args above the shadow window (no return-address clobber). x86 rejects args > `uint.MaxValue`. Argument count bounded by `MaxArguments` (256) to keep frame arithmetic overflow-free. **Byte-level tests only — a live-execution test (5-arg + SSE callee via `CreateRemoteThread`) is still needed to prove the ABI at runtime.**
+- [x] 3.8 Implement x64 stub to pass 3.7. **Deviation (review):** `BuildCallStub` takes `nuint[]` (was `uint[]`). x64 stub is Microsoft-x64-ABI compliant: allocates 32-byte shadow space, keeps 16-byte stack alignment at the inner `call` (frame `K ≡ 8 (mod 16)`, `K ≥ 0x20 + 8·stackArgs`), loads RCX/RDX/R8/R9 with full 64-bit `imm64` (no >4 GiB pointer truncation), and writes stack args above the shadow window (no return-address clobber). x86 rejects args > `uint.MaxValue`. Argument count bounded by `MaxArguments` (256) to keep frame arithmetic overflow-free. **Runtime ABI now proven:** live-execution tests via `CreateRemoteThread` cover 5-arg register+stack delivery (`Execute_sums_register_and_stack_arguments`), 16-byte entry alignment arithmetically (`Execute_delivers_16byte_aligned_stack_to_callee`), and a hardware-alignment-sensitive SSE callee (`Execute_runs_sse_callee_with_five_args`: aligned `movaps` that #GPs unless the stub delivers a 16-byte-aligned stack, combined with a 5th stack arg).
 - [x] 3.9 Confirm no FASM/`ManagedFasm` reference exists in `WhiteMagic` output (assert via a test that scans loaded references)
 
 ## 4. Crash-Safe Execution Slice (spec: remote-execution, function-hooking)
@@ -38,7 +38,7 @@
 - [x] 4.3 Add tests for `DetourManager`/`Detour` in-process: apply redirects, `CallOriginal`, remove restores, named lookup
 - [x] 4.4 Implement `WhiteMagic/Hooking/DetourManager.cs` + `Detour.cs` (inline jmp, x86/x64 form) to pass 4.3
 - [x] 4.5 Add tests for instruction-boundary validation (aligned splice permitted, misaligned rejected when boundary info available)
-- [x] 4.6 Implement minimal prologue length-decoder in `Detour.Apply` to pass 4.5. Default `StubAssembler` covers ONLY the common x86/x64 prologue shapes — enumerate the covered opcodes in code + XML doc (e.g. `push reg` 0x50-0x57, `mov edi,edi` 8B FF, `push ebp`/`mov ebp,esp` 55 8B EC, `sub esp,imm` 83 EC / 81 EC, REX-prefixed forms). On any opcode outside the set, refuse the splice (do not guess). Full arbitrary-prologue validation is gated on the optional Iced backend (task 8.3) — document that slices 2-5 ship partial boundary safety.
+- [x] 4.6 Implement minimal prologue length-decoder in `Detour.Apply` to pass 4.5. Default `StubAssembler` covers ONLY the common x86/x64 prologue shapes — enumerate the covered opcodes in code + XML doc (e.g. `push reg` 0x50-0x57, `mov edi,edi` 8B FF, `push ebp`/`mov ebp,esp` 55 8B EC, `sub esp,imm` 83 EC / 81 EC, REX-prefixed forms). On any opcode outside the set, refuse the splice (do not guess). Full arbitrary-prologue validation is gated on the optional Iced backend (task 8.3) — document that slices 2-5 ship partial boundary safety. **Resolved (8.3):** `DetourManager.PrologueLengthResolver` now accepts `IcedAssembler.GetPrologueLength` for full instruction-boundary validation of arbitrary prologues; the built-in decoder remains the zero-dependency default.
 - [x] 4.7 Add tests for auto-restore: disposing a `MemoryBase` reverts all active patches and detours
 - [x] 4.8 Wire manager registration + `MemoryBase.Dispose` restore to pass 4.7
 - [x] 4.9 Add tests for `MainThreadPump` queue semantics: item runs on hooked thread, result returned, throwing item surfaces exception and pump survives, dispose uninstalls hook (use a self-hosted frame-loop harness in-process)
@@ -66,7 +66,7 @@
 ## 7. High-Level Ergonomics (spec: high-level-api)
 
 - [x] 7.1 Add tests + implement `RemotePointer` indexer (`sharp[addr].Read/Write/Execute` relative to base)
-- [ ] 7.2 Add tests + implement `RemoteModule`/`RemoteFunction` (`sharp["mod"]["fn"]`) resolving export addresses and executing via a chosen strategy
+- [x] 7.2 Add tests + implement `RemoteModule`/`RemoteFunction` (`sharp["mod"]["fn"]`) resolving export addresses and executing via a chosen strategy. **Deviation:** export resolution added to `PeHeaderParser.GetExportAddress` (PE32/PE32+ export directory walk) and **follows export forwarders** (e.g. `kernel32!HeapAlloc` → `NTDLL.RtlAllocateHeap`) into other loaded modules; ordinal forwarders and unresolvable API-set targets throw `NotSupportedException`. `RemoteModule` resolves the base via `Process.Modules` (name match tolerant of `.dll`/case). `RemoteFunction.Execute<T>` defaults to the always-available `RemoteThreadExecutor`; `Address` is exposed for pump routing and `CreateDelegate<T>` for the in-process tier. Tests cross-check resolved addresses against the OS `GetProcAddress` (direct export + forwarder) and execute `kernel32!GetCurrentProcessId` end-to-end.
 - [x] 7.3 Add tests + implement `ManagedPeb`/`ManagedTeb` field reads
 - [x] 7.4 Add tests + implement `WindowFactory`/`RemoteWindow` (enumerate, move/resize/title/activate/flash, query by class)
 - [x] 7.5 Add tests + implement keyboard/mouse simulation (PostMessage + SendInput) to a target window
@@ -75,13 +75,13 @@
 
 ## 8. Optional Iced Backend (spec: managed-assembler)
 
-- [ ] 8.1 Add `Iced` package reference behind an `IcedAssembler : IAssembler` in a way that keeps the default `StubAssembler` dependency-free
+- [x] 8.1 Add `Iced` package reference behind an `IcedAssembler : IAssembler` in a way that keeps the default `StubAssembler` dependency-free. Iced 1.21.0 added to `WhiteMagic.csproj`; only constructing `IcedAssembler` pulls it into a behavioral path. `StubAssembler` never references it.
-- [ ] 8.2 Add tests + implement `IcedAssembler.Assemble(text, origin)` for arbitrary mnemonics and origin-relative encoding
+- [x] 8.2 Add tests + implement `IcedAssembler.Assemble(text, origin)` for arbitrary mnemonics and origin-relative encoding. **Deviation:** Iced ships a *fluent* code assembler and a decoder but **no text parser**, so `Assemble` bridges Intel-syntax text onto Iced's `Assembler` by reflection — the mnemonic selects the matching fluent method and operands bind to registers (reflected from `AssemblerRegisters`), immediates, or labels; origin-relative encoding via `Assembler.Assemble(writer, origin)`. Register/immediate/label operands and label-relative branches are supported; **memory operands (`[reg+disp]`) throw `NotSupportedException`** (a caller needing those emits bytes directly). Tests round-trip via Iced's decoder and assert origin-relative branch targets.
-- [ ] 8.3 Add tests + wire full prologue instruction-boundary validation (D5) using the Iced disassembler when present
+- [x] 8.3 Add tests + wire full prologue instruction-boundary validation (D5) using the Iced disassembler when present. `IcedAssembler.GetPrologueLength` decodes arbitrary instructions via Iced's `Decoder`; `DetourManager.PrologueLengthResolver` (a `PrologueLengthResolver` delegate) defaults to the built-in `PrologueDecoder` and is swappable to the Iced resolver, threaded into each `Detour`. Tests prove Iced resolves a prologue (`mov rax,rcx` = `48 8B C1`) the built-in decoder rejects.
 
 ## 9. Verification
 
 - [x] 9.1 Run full test suite: `dotnet test WhiteMagicTest/WhiteMagicTest.csproj` — all pass (180 pass, 4 integration/interactive skipped)
 - [x] 9.2 Run full build (`dotnet build WhiteMagic.slnx`) — zero errors, zero new warnings in `WhiteMagic`
-- [ ] 9.3 Confirm existing BlackMagic/its tests are unchanged and still green
+- [x] 9.3 Confirm existing BlackMagic/its tests are unchanged and still green. WhiteMagic is a separate, git-ignored project under `reference/` sharing no source or build with BlackMagic; `dotnet test reference/Blackmagic/BlackMagic.slnx` = 17 passing, 0 failing (only pre-existing XML-doc warnings).
-- [ ] 9.4 Update `docs/memory-library-comparison.md` "WhiteMagic — synthesis" section with any deviations discovered during implementation
+- [x] 9.4 Update `docs/memory-library-comparison.md` "WhiteMagic — synthesis" section with any deviations discovered during implementation. Added a "Deviations discovered during implementation" subsection (D1 RPM-on-self, MarshalCache size split, x64 ABI + SSE proof, export forwarders, partial prologue safety, injection bitness, bounds/protection hardening) and corrected the `InProcessReader` line in the architecture diagram.