public void WriteObjectFile(string fileName, BasicObjectFile file)
{
byte[] dataBytes;
// "trick" our object file into writing data into our arrays.
using (var strm = new MemoryStream())
{
foreach (var dataElement in file.DataElements)
{
dataElement.WriteDataToFile(strm);
}
dataBytes = strm.ToArray();
}
byte[] textBytes;
using (var strm = new MemoryStream())
{
foreach (var textElement in file.TextElements)
{
textElement.WriteDataToFile(strm);
}
textBytes = strm.ToArray();
}
using (var underlyingWriter = new ELF_Wrapper.ELF_Writer())
{
underlyingWriter.AddDataSection(dataBytes, Common.CommonConstants.BASE_DATA_ADDRESS);
underlyingWriter.AddTextSection(textBytes, Common.CommonConstants.BASE_TEXT_ADDRESS);
underlyingWriter.AddSymbolTable(file.SymbolTable);
underlyingWriter.WriteFile(fileName);
}
}
/// <summary>
/// Tries to handle any linkage declarations
/// </summary>
/// <param name="trimmedLine">The line with leading/trailing whitespace removed.</param>
/// <param name="lineNum">The current line number.</param>
/// <param name="objFile">The basic object file that will be written to.</param>
/// <returns>Returns true if a linkage directive was processed in this line. Otherwise, returns false.</returns>
private bool TryHandlingLinkageDeclaration(string trimmedLine, int lineNum, BasicObjectFile objFile)
{
// tokenize the line;
string[] tokens = trimmedLine.Split(' ');
bool isLinkageDec = false;
if (IsLinkageDeclaration(tokens[0]))
{
isLinkageDec = true;
// we expect three tokens,
if (tokens[0] == ".extern")
{
int declarationSize = 0;
if (tokens.Length != 3)
{
throw new AssemblyException(lineNum, "Expected symbol name and byte size declaration after .extern token.");
}
else if (!int.TryParse(tokens[2], out declarationSize))
{
throw new AssemblyException(lineNum, ".extern requires a non-negative 32-bit integer size.");
}
else if (declarationSize < 0)
{
throw new AssemblyException(lineNum, ".extern requires a non-negative 32-bit integer size.");
}
else
{
objFile.AddExternElement(declarationSize);
}
}
}
return(isLinkageDec);
}
/// <summary>
/// Adds a non-trivial data size element to the object file.
/// </summary>
/// <param name="objFile">The object file to add the element to.</param>
/// <param name="dataType">The string token declaring the data type</param>
/// <param name="elemValue">The string</param>
/// <param name="lineNum"></param>
private void AddNonTrivialDataElementToObjectFile(BasicObjectFile objFile, string dataType, string elemValue)
{
if (dataType == ".ascii")
{
objFile.AddAsciiString(elemValue);
}
else if (dataType == ".asciiz")
{
objFile.AddNullTerminatedAsciiString(elemValue);
}
else if (dataType == ".space")
{
int numReservedSpace = ParserCommon.DetermineNonTrivialDataLength(dataType, elemValue);
for (int i = 0; i < numReservedSpace; ++i)
{
objFile.AddDataElement((byte)0);
}
}
else
{
throw new ArgumentException("Unknown data type " + dataType + " passed as non-trivial data type.");
}
}
/// <summary>
/// Outputs all data in the BasicObjectFile to the specified format.
/// </summary>
/// <param name="fileName">The file path to generate the output at.</param>
/// <param name="file">The data that will be written to the file.</param>
public void WriteObjectFile(string fileName, BasicObjectFile file)
{
using (FileStream fs = File.Open(fileName, FileMode.Create))
{
using (MemoryStream tmpStrm = new MemoryStream())
{
// write the .data segment metadata
SegmentData dataMdataInfo;
dataMdataInfo.SegmentName = ".dmdta";
dataMdataInfo.StartingOffset = CalculateSegmentOffset(tmpStrm.Position);
dataMdataInfo.SegmentSize = WriteMetadataToFile(tmpStrm, file.DataElements);
// write the actual .data segment
SegmentData dataInfo;
dataInfo.SegmentName = ".data";
dataInfo.StartingOffset = CalculateSegmentOffset(tmpStrm.Position);
dataInfo.SegmentSize = WriteDataToFile(tmpStrm, file.DataElements);
SegmentData textInfo;
textInfo.SegmentName = ".text";
textInfo.StartingOffset = CalculateSegmentOffset(tmpStrm.Position);
textInfo.SegmentSize = WriteDataToFile(tmpStrm, file.TextElements);
// write the .extern segment.
SegmentData externInfo;
externInfo.SegmentName = ".extern";
externInfo.StartingOffset = CalculateSegmentOffset(tmpStrm.Position);
externInfo.SegmentSize = WriteDataToFile(tmpStrm, file.ExternElements);
// write the symbol table
SegmentData symInfo;
symInfo.SegmentName = ".symtbl";
symInfo.StartingOffset = CalculateSegmentOffset(tmpStrm.Position);
symInfo.SegmentSize = WriteSymbolTableToFile(tmpStrm, file.SymbolTable);
// write the source map table segment.
SegmentData srcMapInfo;
srcMapInfo.SegmentName = ".srcmap";
srcMapInfo.StartingOffset = CalculateSegmentOffset(tmpStrm.Position);
srcMapInfo.SegmentSize = WriteSourceMappingInfoToFile(tmpStrm, file.DebugData);
// add the various sizes.
// the order matters, as this is the order in which they are written to the header.
var sizeList = new List <SegmentData>();
sizeList.Add(dataMdataInfo);
sizeList.Add(dataInfo);
sizeList.Add(textInfo);
sizeList.Add(srcMapInfo);
sizeList.Add(externInfo);
sizeList.Add(symInfo);
// write the actual file header, now that we know our absolute offsets
WriteHeader(fs, sizeList);
// copy the temp stream to the actual file stream.
tmpStrm.Seek(0, SeekOrigin.Begin);
tmpStrm.CopyTo(fs);
}
}
}
/// <summary>
/// Adds a half element to the object file.
/// </summary>
/// <param name="objFile">The object file to add to.</param>
/// <param name="fullText">The full text of the assembly line.</param>
/// <param name="declarationToken">The token specifying the declaration of the size parameter.</param>
private void AddShortElementToFile(BasicObjectFile objFile, string fullText, string declarationToken)
{
// find the token directly after the size directive
int substrBeginIdx = fullText.IndexOf(declarationToken) + declarationToken.Length;
string arguments = fullText.Substring(substrBeginIdx);
// split by commas.
string[] tokenizedArgs = arguments.Split(new[] { ',' });
tokenizedArgs = tokenizedArgs.Apply((str) => str.Trim()).ToArray();
// iterate through each element in the "array".
foreach (string token in tokenizedArgs)
{
// if it contains a ':' character, then this itself is an array of the initialized token.
if (token.Contains(':'))
{
string[] subtokens = token.Split(new[] { ':' }).Apply((str) => str.Trim()).ToArray();
if (subtokens.Length == 2)
{
int numElems = int.Parse(subtokens[1]);
// this syntax is wonky; we're trying to parse literal char elements
// as well as normal bytes here.
if (!IntExtensions.TryParseEx(subtokens[0], out short elemToAdd))
{
// see if we can resolve the string as a symbol.
Symbol sym = m_SymTbl.GetSymbol(subtokens[0]);
elemToAdd = (short)sym.Address;
}
for (int i = 0; i < numElems; ++i)
{
objFile.AddDataElement(elemToAdd);
}
}
else
{
throw new ArgumentException("Expected size parameter after ':' token.");
}
}
else
{
// otherwise, it should just be an element (without any size modifiers).
// just parse it and add it.
if (!IntExtensions.TryParseEx(token, out short elemToAdd))
{
// see if we can resolve the string as a symbol.
Symbol sym = m_SymTbl.GetSymbol(token);
elemToAdd = (short)sym.Address;
}
objFile.AddDataElement(elemToAdd);
}
}
}
/// <summary>
/// Adds either one or multiple trivially sized data elements to the object file.
/// </summary>
/// <param name="objFile">The object file to add to.</param>
/// <param name="dataSize">The data size of the element to add.</param>
/// <param name="fullText">The full text of the assembly line.</param>
/// <param name="declarationToken">The token declaring the size of the data.</param>
private void AddTrivialDataElementsToFile(BasicObjectFile objFile, int dataSize, string fullText, string declarationToken)
{
const int BYTE_DATA_SIZE = 1;
const int SHORT_DATA_SIZE = 2;
const int WORD_DATA_SIZE = 4;
const int DWORD_DATA_SIZE = 8;
switch (dataSize)
{
case BYTE_DATA_SIZE:
{
AddByteElementToFile(objFile, fullText, declarationToken);
break;
}
case SHORT_DATA_SIZE:
{
AddShortElementToFile(objFile, fullText, declarationToken);
break;
}
case WORD_DATA_SIZE:
{
AddIntElementToFile(objFile, fullText, declarationToken);
break;
}
case DWORD_DATA_SIZE:
{
AddLongElementToFile(objFile, fullText, declarationToken);
break;
}
default:
{
throw new ArgumentException("Unknown data size passed to AddTrivialDataElementToObjectFile");
}
}
}
/// <summary>
/// Generates code for a .text instruction in the file.
/// </summary>
/// <param name="fileName">The source file name.</param>
/// <param name="asmLine">The line to parse.</param>
/// <param name="objFile">The object file that will be written to.</param>
/// <param name="currAlignment">The current specified alignment of the file. Unused for .text parsers.</param>
public void GenerateCodeForSegment(string fileName, LineData asmLine, BasicObjectFile objFile, int currAlignment)
{
// scan to the first instruction.
// this could share the same line as a label, so split on ':' and ','
string[] tokenizedStr = asmLine.Text.Split(new char[] { ',', ':', ' ' }, StringSplitOptions.RemoveEmptyEntries);
bool foundInstruction = false;
string instructionToken = string.Empty;
for (int i = 0; i < tokenizedStr.Length && !foundInstruction; ++i)
{
string token = tokenizedStr[i].Trim();
// we found our instruction. build a string from this token
// to the end of the array.
if (m_ParserFac.IsInstruction(token))
{
foundInstruction = true;
instructionToken = token;
}
}
if (foundInstruction)
{
// first, validate that the instruction is not the last token in the string.
// try to parse the instruction parameters
// get the substring starting at the index of the next character after the instruction
string instSubstring = asmLine.Text.Substring(asmLine.Text.IndexOf(instructionToken) + instructionToken.Length);
//split the substring at the comma to get the instruction parameters.
string[] argTokens = instSubstring.Split(new[] { ',' }, StringSplitOptions.RemoveEmptyEntries);
// trim whitespace from the beginning and end of each token.
argTokens = argTokens.Apply((str) => str.Trim()).ToArray();
// find the parser for the instruction.
IInstructionGenerator parser = m_ParserFac.GetProcessorForInstruction(instructionToken);
// beq instructions should (hopefully) not generate multiple instructions..
IEnumerable <int> generatedInstructions = parser.GenerateCodeForInstruction(m_CurrTextAddress, argTokens);
var srcInfo = new SourceLineInformation(fileName, asmLine.LineNum, m_CurrTextAddress, asmLine.Text);
objFile.AddSourceInformation(srcInfo);
foreach (int generatedInstruction in generatedInstructions)
{
objFile.AddInstruction(generatedInstruction);
m_CurrTextAddress += CommonConstants.BASE_INSTRUCTION_SIZE_BYTES;
}
}
else
{
// if not an instruction (may be a symbol)
// preprocesor instruction shouldn't bring us here.
// make sure the user is not typing garbage.
string symStr = tokenizedStr[0];
if (!m_SymTbl.ContainsSymbol(symStr))
{
throw new AssemblyException(asmLine.LineNum, "Unknown instruction \"" + asmLine.Text + "\" found.");
}
}
}
/// <summary>
/// Generates code for the given assembly file.
/// </summary>
/// <param name="reader">A StreamReader instance that will read the input assembly file.</param>
/// <param name="objFile">The basic object file that will be written to.</param>
public void GenerateCode(string fileName, StreamReader reader, BasicObjectFile objFile)
{
SegmentType currSegmentType = SegmentType.Invalid;
int currAlignment = CommonConstants.DEFAULT_ALIGNMENT;
// a list of all exceptions we encounter during parsing.
// users can view them all at once instead of working through them piecemeal.
var exceptionList = new List <AssemblyException>();
int lineNum = 0;
while (!reader.EndOfStream)
{
try
{
++lineNum;
// trim the whitespace from any read-in line.
string line = reader.ReadLine().Trim();
// get a substring up until the commented line, ignoring those in user quotes.
line = ParserCommon.GetSanitizedString(line);
// ignore blank lines. trim should remove all whitespace
if (line.Any())
{
LineParseResults directiveResults = ParserCommon.HandleAssemblerDirective(line, lineNum, currSegmentType, currAlignment);
// if we have a new data alignment, we need to record this in case the disassembler
// parses this file. otherwise, it will interpret padding bytes as actual data elements.
if (currAlignment != directiveResults.NewAlignment)
{
objFile.AddAlignmentChangeDeclaration(directiveResults.NewAlignment);
}
currAlignment = directiveResults.NewAlignment;
currSegmentType = directiveResults.NewSegment;
// if our segment type is valid, then we're processing actual data versus an assembler directive.
if (!directiveResults.IsLineAssemblerDirective &&
currSegmentType != SegmentType.Invalid)
{
if (!TryHandlingLinkageDeclaration(line, lineNum, objFile))
{
ISegmentCodeGenerator codeGen = m_CodeGenFac.GetCodeGeneratorForSegment(currSegmentType);
var asmLine = new LineData(line, lineNum);
try
{
codeGen.GenerateCodeForSegment(fileName, asmLine, objFile, currAlignment);
}
catch (AssemblyException)
{
throw;
}
catch (Exception ex)
{
throw new AssemblyException(lineNum, ex.Message);
}
}
}
}
}
catch (AssemblyException ex)
{
exceptionList.Add(ex);
}
}
// if any exceptions were encountered, throw an aggregate exception with
// all of the encountered exceptions.
if (exceptionList.Any())
{
throw new AggregateAssemblyError(exceptionList);
}
// reset the StreamReader to the beginning position.
reader.Seek(0, SeekOrigin.Begin);
}
/// <summary>
/// Task for assembling one individual file.
/// </summary>
/// <param name="inputFile">The input file to assemble.</param>
/// <param name="logger">The logging implementation to log errors/info to.</param>
/// <param name="options">The options to use while assembling.</param>
/// <returns>True if the assembler could successfully generate code for the file; otherwise returns false.</returns>
public AssemblerResult AssembleFile(string inputFile, string outputFile, ILogger logger, AssemblerOptions options)
{
var result = new AssemblerResult();
logger.Log(LogLevel.Info, "Invoking assembler for file " + inputFile);
try
{
bool furtherProcessingNeeded = true;
if (File.Exists(inputFile) &&
File.Exists(outputFile))
{
DateTime inputFileWriteTime = File.GetLastWriteTimeUtc(inputFile);
DateTime outputFileWriteTime = File.GetLastWriteTimeUtc(outputFile);
if (outputFileWriteTime > inputFileWriteTime)
{
logger.Log(LogLevel.Info, "Nothing to do for file " + inputFile);
furtherProcessingNeeded = false;
}
}
if (furtherProcessingNeeded)
{
using (var reader = new StreamReader(File.OpenRead(inputFile)))
{
var symTable = new SymbolTable();
// build the symbol table
var instructionProcFac = new InstructionProcessorFactory(symTable);
var symTableBuilder = new SymbolTableBuilder(logger, instructionProcFac);
symTableBuilder.GenerateSymbolTableForSegment(reader, SegmentType.Data, symTable);
symTableBuilder.GenerateSymbolTableForSegment(reader, SegmentType.Text, symTable);
// use the symbol table to generate code with references resolved.
var objFile = new BasicObjectFile(inputFile, symTable);
var codeGenerator = new CodeGenerator(logger, symTable, instructionProcFac);
codeGenerator.GenerateCode(inputFile, reader, objFile);
if (!objFile.TextElements.Any())
{
logger.Log(LogLevel.Warning, "No .text segment to assemble. Stop.");
result.OperationSuccessful = false;
}
else
{
// write the object file out.
var writerFac = new ObjectFileWriterFactory();
IObjectFileWriter writer = writerFac.GetWriterForOutputFile(outputFile);
writer.WriteObjectFile(outputFile, objFile);
}
}
}
}
catch (AggregateAssemblyError ex)
{
foreach (AssemblyException asEx in ex.AssemblyErrors)
{
logger.Log(LogLevel.Critical, "In file \"" + inputFile + "\" (line " + asEx.LineNumber + "):\n\t");
logger.Log(LogLevel.Critical, asEx.Message);
result.AddUserAssemblyError(asEx);
}
}
catch (Exception ex)
{
logger.Log(LogLevel.Critical, ex.StackTrace);
logger.Log(LogLevel.Critical, ex.Message);
if (ex.InnerException != null)
{
logger.Log(LogLevel.Critical, ex.InnerException.StackTrace);
logger.Log(LogLevel.Critical, ex.InnerException.Message);
}
result.AddInternalAssemblerError(ex);
}
return(result);
}
/// <summary>
/// Generates the byte representation of an instruction from a line of assembly code.
/// </summary>
/// <param name="asmLine">The line to parse.</param>
/// <param name="objFile">The object file that will be written to.</param>
/// <param name="currAlignment">The current specified alignment of the file.</param>
public void GenerateCodeForSegment(string fileName, LineData asmLine, BasicObjectFile objFile, int currAlignment)
{
string[] tokens = asmLine.Text.Split(' ', '\t');
string[] fixedTokens = ParserCommon.GetTrimmedTokenArray(tokens).ToArray();
bool foundDataDeclaration = false;
int dataDeclarationIdx = 0;
for (int i = 0; i < fixedTokens.Length && !foundDataDeclaration; ++i)
{
if (ParserCommon.IsDataDeclaration(fixedTokens[i]))
{
foundDataDeclaration = true;
dataDeclarationIdx = i;
}
}
// we found a data declaration; make sure that there's at least one value following it.
if (foundDataDeclaration)
{
if (dataDeclarationIdx + 1 < fixedTokens.Length)
{
// if it is a trivial type, use our precomputed map to get the size.
if (ParserCommon.IsTrivialDataType(fixedTokens[dataDeclarationIdx]))
{
// determine before writing the next data element if we need to add padding.
int paddingSize = ParserCommon.GetNumPaddingBytes(m_NumBytesLaidOut, currAlignment);
int dataSize = ParserCommon.DetermineTrivialDataSize(fixedTokens[dataDeclarationIdx]);
int numElements = ParserCommon.GetArraySize(asmLine.Text, fixedTokens[dataDeclarationIdx]);
if (dataSize > paddingSize)
{
// add as much padding as we need to reach the next alignment boundary.
for (int i = 0; i < paddingSize; ++i)
{
objFile.AddDataElement((byte)0);
m_NumBytesLaidOut += 1;
}
}
int totalReservedSize = dataSize * numElements;
m_NumBytesLaidOut += totalReservedSize;
AddTrivialDataElementsToFile(objFile, dataSize, asmLine.Text, fixedTokens[dataDeclarationIdx]);
}
// see if we can figure out the string length
else if (ParserCommon.IsStringDeclaration(fixedTokens[dataDeclarationIdx]))
{
int paddingSize = ParserCommon.GetNumPaddingBytes(m_NumBytesLaidOut, currAlignment);
// if this is a string declaration, then get the original string data
string dataStr = ParserCommon.GetStringData(asmLine.Text);
int dataSize = ParserCommon.DetermineNonTrivialDataLength(fixedTokens[dataDeclarationIdx], dataStr);
if (dataSize > paddingSize)
{
// add as much padding as we need to reach the next alignment boundary.
for (int i = 0; i < paddingSize; ++i)
{
objFile.AddDataElement((byte)0);
m_NumBytesLaidOut += 1;
}
}
// add the string data to the object file.
AddNonTrivialDataElementToObjectFile(objFile, fixedTokens[dataDeclarationIdx], dataStr);
m_NumBytesLaidOut += dataSize;
}
// otherwise, this must be a .space declaration. just get the size following it.
else
{
int dataSize = ParserCommon.DetermineNonTrivialDataLength(fixedTokens[dataDeclarationIdx], fixedTokens[dataDeclarationIdx + 1]);
int paddingSize = ParserCommon.GetNumPaddingBytes(dataSize, currAlignment);
// fill the space and padding with zeroes.
for (int i = 0; i < dataSize + paddingSize; ++i)
{
objFile.AddDataElement((byte)0);
}
// we expect one token after this word.
// otherwise, it may be garbage that we should detect.
if (fixedTokens.Length > dataDeclarationIdx + 2)
{
throw new AssemblyException(asmLine.LineNum, "Unknown token \"" + fixedTokens[dataDeclarationIdx + 2] + "\" found.");
}
}
}
else
{
throw new AssemblyException(asmLine.LineNum, "Expected data value after token " + fixedTokens[dataDeclarationIdx]);
}
}
// check to see if this is just a label.
// otherwise, it is probably garbage that we should throw.
else if (!ParserCommon.ContainsLabel(asmLine.Text))
{
throw new AssemblyException(asmLine.LineNum, "Unable to ascertain data type from line " + asmLine.Text);
}
}
/// <summary>
/// Adds a byte element to the object file.
/// </summary>
/// <param name="objFile">The object file to add to.</param>
/// <param name="fullText">The full text of the assembly line.</param>
/// <param name="declarationToken">The token specifying the declaration of the size parameter.</param>
private void AddByteElementToFile(BasicObjectFile objFile, string fullText, string declarationToken)
{
// find the token directly after the size directive
int substrBeginIdx = fullText.IndexOf(declarationToken) + declarationToken.Length;
string arguments = fullText.Substring(substrBeginIdx);
// split by commas.
string[] tokenizedArgs = arguments.Split(new[] { ',' });
tokenizedArgs = tokenizedArgs.Apply((str) => str.Trim()).ToArray();
// iterate through each element in the "array".
foreach (string token in tokenizedArgs)
{
// if it contains a ':' character, then this itself is an array of the initialized token.
if (token.Contains(':'))
{
string[] subtokens = token.Split(new[] { ':' }).Apply((str) => str.Trim()).ToArray();
if (subtokens.Length == 2)
{
int numElems = int.Parse(subtokens[1]);
// first, try to get the value as a byte.
if (!IntExtensions.TryParseEx(subtokens[0], out byte byteElem))
{
// if we fail, then try parsing the character as a literal.
if (!StringUtils.TryParseCharacterLiteralAsByte(subtokens[0], out byteElem))
{
// as a fallback, see if we can resolve the string as a symbol.
Symbol sym = m_SymTbl.GetSymbol(subtokens[0]);
byteElem = (byte)sym.Address;
}
}
for (int i = 0; i < numElems; ++i)
{
objFile.AddDataElement(byteElem);
}
}
else
{
throw new ArgumentException("Expected size parameter after ':' token.");
}
}
else
{
// otherwise, it should just be an element (without any size modifiers).
// just parse it and add it.
// first, try to get the value as a byte.
if (!IntExtensions.TryParseEx(token, out byte byteElem))
{
// if we fail, then try parsing the character as a literal.
if (!StringUtils.TryParseCharacterLiteralAsByte(token, out byteElem))
{
// as a fallback, see if we can resolve the string as a symbol.
Symbol sym = m_SymTbl.GetSymbol(token);
byteElem = (byte)sym.Address;
}
}
objFile.AddDataElement(byteElem);
}
}
}