1
0
mirror of https://github.com/fadden/6502bench.git synced 2024-12-12 05:31:04 +00:00
6502bench/SourceGen/AsmGen/AsmMerlin32.cs

903 lines
38 KiB
C#
Raw Normal View History

/*
* Copyright 2019 faddenSoft
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
using System;
using System.Collections.Generic;
using System.ComponentModel;
using System.Diagnostics;
using System.IO;
using System.Text;
using Asm65;
using CommonUtil;
namespace SourceGen.AsmGen {
#region IGenerator
/// <summary>
/// Generate source code compatible with Brutal Deluxe's Merlin 32 assembler
/// (https://www.brutaldeluxe.fr/products/crossdevtools/merlin/).
/// </summary>
public class GenMerlin32 : IGenerator {
private const string ASM_FILE_SUFFIX = "_merlin32.S"; // must start with underscore
// IGenerator
public DisasmProject Project { get; private set; }
// IGenerator
public Formatter SourceFormatter { get; private set; }
// IGenerator
public AppSettings Settings { get; private set; }
// IGenerator
public AssemblerQuirks Quirks { get; private set; }
// IGenerator
public LabelLocalizer Localizer { get { return mLocalizer; } }
// IGenerator
public int StartOffset { get { return 0; } }
/// <summary>
/// Working directory, i.e. where we write our output file(s).
/// </summary>
private string mWorkDirectory;
/// <summary>
/// If set, long labels get their own line.
/// </summary>
private bool mLongLabelNewLine;
/// <summary>
/// Output column widths.
/// </summary>
private int[] mColumnWidths;
/// <summary>
/// Base filename. Typically the project file name without the ".dis65" extension.
/// </summary>
private string mFileNameBase;
/// <summary>
/// StringBuilder to use when composing a line. Held here to reduce allocations.
/// </summary>
private StringBuilder mLineBuilder = new StringBuilder(100);
/// <summary>
/// Label localization helper.
/// </summary>
private LabelLocalizer mLocalizer;
/// <summary>
/// Stream to send the output to.
/// </summary>
private StreamWriter mOutStream;
ORG rework, part 6 Added support for non-addressable regions, which are useful for things like file headers stripped out by the system loader, or chunks that get loaded into non-addressable graphics RAM. Regions are specified with the "NA" address value. The code list displays the address field greyed out, starting from zero (which is kind of handy if you want to know the relative offset within the region). Putting labels in non-addressable regions doesn't make sense, but symbol resolution is complicated enough that we really only have two options: ignore the labels entirely, or allow them but warn of their presence. The problem isn't so much the label, which you could legitimately want to access from an extension script, but rather the references to them from code or data. So we keep the label and add a warning to the Messages list when we see a reference. Moved NON_ADDR constants to Address class. AddressMap now has a copy. This is awkward because Asm65 and CommonUtil don't share. Updated the asm code generators to understand NON_ADDR, and reworked the API so that Merlin and cc65 output is correct for nested regions. Address region changes are now noted in the anattribs array, which makes certain operations faster than checking the address map. It also fixes a failure to recognize mid-instruction region changes in the code analyzer. Tweaked handling of synthetic regions, which are non-addressable areas generated by the linear address map traversal to fill in any "holes". The address region editor now treats attempts to edit them as creation of a new region.
2021-10-01 01:07:21 +00:00
/// <summary>
/// Address of next byte of output.
/// </summary>
private int mNextAddress = -1;
/// <summary>
/// True if we've seen an "is relative" flag in a block of address region start directives.
/// </summary>
/// <remarks>
/// The trick with IsRelative is that, if there are multiple arstarts at the same
/// offset, we need to output some or all of them, starting from the one just before
/// the first IsRelative start. We probably want to disable the use of Flush and
/// just generate them as they appear, using the next Flush as the signal to return
/// to standard behavior.
/// </remarks>
bool mIsInRelative = false;
/// <summary>
/// Holds detected version of configured assembler.
/// </summary>
private CommonUtil.Version mAsmVersion = CommonUtil.Version.NO_VERSION;
// Interesting versions.
private static CommonUtil.Version V1_0 = new CommonUtil.Version(1, 0);
Various improvements The PseudoOpNames class is increasingly being used in situations where mutability is undesirable. This change makes instances immutable, eliminating the Copy() method and adding a constructor that takes a Dictionary. The serialization code now operates on a Dictionary instead of the class properties, but the JSON encoding is identical, so this doesn't invalidate app settings file data. Added an equality test to PseudoOpNames. In LineListGen, don't reset the line list if the names haven't actually changed. Use a table lookup for C64 character conversions. I figure that should be faster than multiple conditionals on a modern x64 system. Fixed a 64tass generator issue where we tried to query project properties in a call that might not have a project available (specifically, getting FormatConfig values out of the generator for use in the "quick set" buttons for Display Format). Fixed a regression test harness issue where, if the assembler reported success but didn't actually generate output, an exception would be thrown that halted the tests. Increased the width of text entry fields on the Pseudo-Op tab of app settings. The previous 8-character limit wasn't wide enough to hold ACME's "!pseudopc". Also, use TrimEnd() to remove trailing spaces (leading spaces are still allowed). In the last couple of months, Win10 started stalling for a fraction of a second when executing assemblers. It doesn't do this every time; mostly it happens if it has been a while since the assembler was run. My guess is this has to do with changes to the built-in malware scanner. Whatever the case, we now change the mouse pointer to a wait cursor while updating the assembler version cache.
2019-08-17 18:14:05 +00:00
// Pseudo-op string constants.
private static PseudoOp.PseudoOpNames sDataOpNames =
new PseudoOp.PseudoOpNames(new Dictionary<string, string> {
{ "EquDirective", "equ" },
{ "VarDirective", "equ" },
{ "ArStartDirective", "org" },
{ "ArEndDirective", "adrend" }, // on-screen display only
Various improvements The PseudoOpNames class is increasingly being used in situations where mutability is undesirable. This change makes instances immutable, eliminating the Copy() method and adding a constructor that takes a Dictionary. The serialization code now operates on a Dictionary instead of the class properties, but the JSON encoding is identical, so this doesn't invalidate app settings file data. Added an equality test to PseudoOpNames. In LineListGen, don't reset the line list if the names haven't actually changed. Use a table lookup for C64 character conversions. I figure that should be faster than multiple conditionals on a modern x64 system. Fixed a 64tass generator issue where we tried to query project properties in a call that might not have a project available (specifically, getting FormatConfig values out of the generator for use in the "quick set" buttons for Display Format). Fixed a regression test harness issue where, if the assembler reported success but didn't actually generate output, an exception would be thrown that halted the tests. Increased the width of text entry fields on the Pseudo-Op tab of app settings. The previous 8-character limit wasn't wide enough to hold ACME's "!pseudopc". Also, use TrimEnd() to remove trailing spaces (leading spaces are still allowed). In the last couple of months, Win10 started stalling for a fraction of a second when executing assemblers. It doesn't do this every time; mostly it happens if it has been a while since the assembler was run. My guess is this has to do with changes to the built-in malware scanner. Whatever the case, we now change the mouse pointer to a wait cursor while updating the assembler version cache.
2019-08-17 18:14:05 +00:00
//RegWidthDirective
//DataBankDirective
Various improvements The PseudoOpNames class is increasingly being used in situations where mutability is undesirable. This change makes instances immutable, eliminating the Copy() method and adding a constructor that takes a Dictionary. The serialization code now operates on a Dictionary instead of the class properties, but the JSON encoding is identical, so this doesn't invalidate app settings file data. Added an equality test to PseudoOpNames. In LineListGen, don't reset the line list if the names haven't actually changed. Use a table lookup for C64 character conversions. I figure that should be faster than multiple conditionals on a modern x64 system. Fixed a 64tass generator issue where we tried to query project properties in a call that might not have a project available (specifically, getting FormatConfig values out of the generator for use in the "quick set" buttons for Display Format). Fixed a regression test harness issue where, if the assembler reported success but didn't actually generate output, an exception would be thrown that halted the tests. Increased the width of text entry fields on the Pseudo-Op tab of app settings. The previous 8-character limit wasn't wide enough to hold ACME's "!pseudopc". Also, use TrimEnd() to remove trailing spaces (leading spaces are still allowed). In the last couple of months, Win10 started stalling for a fraction of a second when executing assemblers. It doesn't do this every time; mostly it happens if it has been a while since the assembler was run. My guess is this has to do with changes to the built-in malware scanner. Whatever the case, we now change the mouse pointer to a wait cursor while updating the assembler version cache.
2019-08-17 18:14:05 +00:00
{ "DefineData1", "dfb" },
{ "DefineData2", "dw" },
{ "DefineData3", "adr" },
{ "DefineData4", "adrl" },
{ "DefineBigData2", "ddb" },
//DefineBigData3
//DefineBigData4
{ "Fill", "ds" },
{ "Dense", "hex" },
{ "Uninit", "ds" },
//Junk
//Align
Various improvements The PseudoOpNames class is increasingly being used in situations where mutability is undesirable. This change makes instances immutable, eliminating the Copy() method and adding a constructor that takes a Dictionary. The serialization code now operates on a Dictionary instead of the class properties, but the JSON encoding is identical, so this doesn't invalidate app settings file data. Added an equality test to PseudoOpNames. In LineListGen, don't reset the line list if the names haven't actually changed. Use a table lookup for C64 character conversions. I figure that should be faster than multiple conditionals on a modern x64 system. Fixed a 64tass generator issue where we tried to query project properties in a call that might not have a project available (specifically, getting FormatConfig values out of the generator for use in the "quick set" buttons for Display Format). Fixed a regression test harness issue where, if the assembler reported success but didn't actually generate output, an exception would be thrown that halted the tests. Increased the width of text entry fields on the Pseudo-Op tab of app settings. The previous 8-character limit wasn't wide enough to hold ACME's "!pseudopc". Also, use TrimEnd() to remove trailing spaces (leading spaces are still allowed). In the last couple of months, Win10 started stalling for a fraction of a second when executing assemblers. It doesn't do this every time; mostly it happens if it has been a while since the assembler was run. My guess is this has to do with changes to the built-in malware scanner. Whatever the case, we now change the mouse pointer to a wait cursor while updating the assembler version cache.
2019-08-17 18:14:05 +00:00
{ "StrGeneric", "asc" },
{ "StrReverse", "rev" },
//StrNullTerm
{ "StrLen8", "str" },
{ "StrLen16", "strl" },
{ "StrDci", "dci" },
});
private const string REG_WIDTH_DIRECTIVE = "mx";
// IGenerator
public void GetDefaultDisplayFormat(out PseudoOp.PseudoOpNames pseudoOps,
out Formatter.FormatConfig formatConfig) {
Various improvements The PseudoOpNames class is increasingly being used in situations where mutability is undesirable. This change makes instances immutable, eliminating the Copy() method and adding a constructor that takes a Dictionary. The serialization code now operates on a Dictionary instead of the class properties, but the JSON encoding is identical, so this doesn't invalidate app settings file data. Added an equality test to PseudoOpNames. In LineListGen, don't reset the line list if the names haven't actually changed. Use a table lookup for C64 character conversions. I figure that should be faster than multiple conditionals on a modern x64 system. Fixed a 64tass generator issue where we tried to query project properties in a call that might not have a project available (specifically, getting FormatConfig values out of the generator for use in the "quick set" buttons for Display Format). Fixed a regression test harness issue where, if the assembler reported success but didn't actually generate output, an exception would be thrown that halted the tests. Increased the width of text entry fields on the Pseudo-Op tab of app settings. The previous 8-character limit wasn't wide enough to hold ACME's "!pseudopc". Also, use TrimEnd() to remove trailing spaces (leading spaces are still allowed). In the last couple of months, Win10 started stalling for a fraction of a second when executing assemblers. It doesn't do this every time; mostly it happens if it has been a while since the assembler was run. My guess is this has to do with changes to the built-in malware scanner. Whatever the case, we now change the mouse pointer to a wait cursor while updating the assembler version cache.
2019-08-17 18:14:05 +00:00
pseudoOps = sDataOpNames;
formatConfig = new Formatter.FormatConfig();
SetFormatConfigValues(ref formatConfig);
}
// IGenerator
public void Configure(DisasmProject project, string workDirectory, string fileNameBase,
AssemblerVersion asmVersion, AppSettings settings) {
Debug.Assert(project != null);
Debug.Assert(!string.IsNullOrEmpty(workDirectory));
Debug.Assert(!string.IsNullOrEmpty(fileNameBase));
Project = project;
Quirks = new AssemblerQuirks();
if (asmVersion != null) {
mAsmVersion = asmVersion.Version; // Use the actual version.
} else {
mAsmVersion = V1_0; // No assembler installed, use default.
}
Quirks.NoPcRelBankWrap = true;
Quirks.TracksSepRepNotEmu = true;
mWorkDirectory = workDirectory;
mFileNameBase = fileNameBase;
Settings = settings;
mLongLabelNewLine = Settings.GetBool(AppSettings.SRCGEN_LONG_LABEL_NEW_LINE, false);
AssemblerConfig config = AssemblerConfig.GetConfig(settings,
AssemblerInfo.Id.Merlin32);
mColumnWidths = (int[])config.ColumnWidths.Clone();
}
/// <summary>
/// Configures the assembler-specific format items.
/// </summary>
private void SetFormatConfigValues(ref Formatter.FormatConfig config) {
config.mOperandWrapLen = 64;
config.mForceDirectOpcodeSuffix = string.Empty;
config.mForceAbsOpcodeSuffix = ":";
config.mForceLongOpcodeSuffix = "l";
config.mForceDirectOperandPrefix = string.Empty;
config.mForceAbsOperandPrefix = string.Empty;
config.mForceLongOperandPrefix = string.Empty;
config.mLocalVariableLabelPrefix = "]";
config.mEndOfLineCommentDelimiter = ";";
config.mFullLineCommentDelimiterBase = ";";
config.mBoxLineCommentDelimiter = string.Empty;
config.mNonUniqueLabelPrefix = ":";
config.mCommaSeparatedDense = false;
config.mExpressionMode = Formatter.FormatConfig.ExpressionMode.Merlin;
Formatter.DelimiterSet charSet = new Formatter.DelimiterSet();
charSet.Set(CharEncoding.Encoding.Ascii, Formatter.SINGLE_QUOTE_DELIM);
charSet.Set(CharEncoding.Encoding.HighAscii, Formatter.DOUBLE_QUOTE_DELIM);
config.mCharDelimiters = charSet;
}
// IGenerator; executes on background thread
public GenerationResults GenerateSource(BackgroundWorker worker) {
List<string> pathNames = new List<string>(1);
string fileName = mFileNameBase + ASM_FILE_SUFFIX;
string pathName = Path.Combine(mWorkDirectory, fileName);
pathNames.Add(pathName);
Formatter.FormatConfig config = new Formatter.FormatConfig();
GenCommon.ConfigureFormatterFromSettings(Settings, ref config);
SetFormatConfigValues(ref config);
SourceFormatter = new Formatter(config);
string msg = string.Format(Res.Strings.PROGRESS_GENERATING_FMT, pathName);
worker.ReportProgress(0, msg);
mLocalizer = new LabelLocalizer(Project);
mLocalizer.LocalPrefix = ":";
// don't need to set QuirkNoOpcodeMnemonics
mLocalizer.Analyze();
// Use UTF-8 encoding, without a byte-order mark.
using (StreamWriter sw = new StreamWriter(pathName, false, new UTF8Encoding(false))) {
mOutStream = sw;
if (Settings.GetBool(AppSettings.SRCGEN_ADD_IDENT_COMMENT, false)) {
// No version-specific stuff yet. We're generating code for v1.0.
OutputLine(SourceFormatter.FullLineCommentDelimiter +
string.Format(Res.Strings.GENERATED_FOR_VERSION_FMT,
"Merlin 32", mAsmVersion, string.Empty));
}
GenCommon.Generate(this, sw, worker);
}
mOutStream = null;
return new GenerationResults(pathNames, string.Empty);
}
// IGenerator
public void OutputDataOp(int offset) {
Formatter formatter = SourceFormatter;
byte[] data = Project.FileData;
Anattrib attr = Project.GetAnattrib(offset);
string labelStr = string.Empty;
if (attr.Symbol != null) {
labelStr = mLocalizer.ConvLabel(attr.Symbol.Label);
}
string commentStr = SourceFormatter.FormatEolComment(Project.Comments[offset]);
string opcodeStr, operandStr;
FormatDescriptor dfd = attr.DataDescriptor;
Debug.Assert(dfd != null);
int length = dfd.Length;
Debug.Assert(length > 0);
bool multiLine = false;
switch (dfd.FormatType) {
case FormatDescriptor.Type.Default:
if (length != 1) {
Debug.Assert(false);
length = 1;
}
opcodeStr = sDataOpNames.DefineData1;
int operand = RawData.GetWord(data, offset, length, false);
operandStr = formatter.FormatHexValue(operand, length * 2);
break;
case FormatDescriptor.Type.NumericLE:
opcodeStr = sDataOpNames.GetDefineData(length);
operand = RawData.GetWord(data, offset, length, false);
if (length == 1 && dfd.IsStringOrCharacter &&
((operand & 0x7f) == '{' || (operand & 0x7f) == '}')) {
// Merlin32 can't handle "DFB '{'", so just output hex.
operandStr = formatter.FormatHexValue(operand, length * 2);
} else {
operandStr = PseudoOp.FormatNumericOperand(formatter, Project.SymbolTable,
mLocalizer.LabelMap, dfd, operand, length,
PseudoOp.FormatNumericOpFlags.OmitLabelPrefixSuffix);
}
break;
case FormatDescriptor.Type.NumericBE:
opcodeStr = sDataOpNames.GetDefineBigData(length);
Various improvements The PseudoOpNames class is increasingly being used in situations where mutability is undesirable. This change makes instances immutable, eliminating the Copy() method and adding a constructor that takes a Dictionary. The serialization code now operates on a Dictionary instead of the class properties, but the JSON encoding is identical, so this doesn't invalidate app settings file data. Added an equality test to PseudoOpNames. In LineListGen, don't reset the line list if the names haven't actually changed. Use a table lookup for C64 character conversions. I figure that should be faster than multiple conditionals on a modern x64 system. Fixed a 64tass generator issue where we tried to query project properties in a call that might not have a project available (specifically, getting FormatConfig values out of the generator for use in the "quick set" buttons for Display Format). Fixed a regression test harness issue where, if the assembler reported success but didn't actually generate output, an exception would be thrown that halted the tests. Increased the width of text entry fields on the Pseudo-Op tab of app settings. The previous 8-character limit wasn't wide enough to hold ACME's "!pseudopc". Also, use TrimEnd() to remove trailing spaces (leading spaces are still allowed). In the last couple of months, Win10 started stalling for a fraction of a second when executing assemblers. It doesn't do this every time; mostly it happens if it has been a while since the assembler was run. My guess is this has to do with changes to the built-in malware scanner. Whatever the case, we now change the mouse pointer to a wait cursor while updating the assembler version cache.
2019-08-17 18:14:05 +00:00
if ((string.IsNullOrEmpty(opcodeStr))) {
// Nothing defined, output as comma-separated single-byte values.
GenerateShortSequence(offset, length, out opcodeStr, out operandStr);
} else {
operand = RawData.GetWord(data, offset, length, true);
operandStr = PseudoOp.FormatNumericOperand(formatter, Project.SymbolTable,
mLocalizer.LabelMap, dfd, operand, length,
PseudoOp.FormatNumericOpFlags.OmitLabelPrefixSuffix);
}
break;
case FormatDescriptor.Type.Fill:
opcodeStr = sDataOpNames.Fill;
if (data[offset] == 0) {
operandStr = length.ToString();
} else {
operandStr = length + "," + formatter.FormatHexValue(data[offset], 2);
}
break;
case FormatDescriptor.Type.Dense:
multiLine = true;
opcodeStr = operandStr = null;
OutputDenseHex(offset, length, labelStr, commentStr);
break;
case FormatDescriptor.Type.Uninit:
case FormatDescriptor.Type.Junk:
int fillVal = Helper.CheckRangeHoldsSingleValue(data, offset, length);
if (fillVal >= 0) {
opcodeStr = sDataOpNames.Fill;
if (dfd.FormatSubType == FormatDescriptor.SubType.Align256 &&
GenCommon.CheckJunkAlign(offset, dfd, Project.AddrMap)) {
// special syntax for page alignment
if (fillVal == 0) {
operandStr = "\\";
} else {
operandStr = "\\," + formatter.FormatHexValue(fillVal, 2);
}
} else if (length == 1 && fillVal != 0x00) {
// Single-byte HEX looks better than "ds 1,$xx", and will match up
// with adjacent multi-byte junk/uninit.
multiLine = true;
opcodeStr = operandStr = null;
OutputDenseHex(offset, length, labelStr, commentStr);
} else {
if (fillVal == 0) {
operandStr = length.ToString();
} else {
operandStr = length + "," + formatter.FormatHexValue(fillVal, 2);
}
}
} else {
// treat same as Dense
multiLine = true;
opcodeStr = operandStr = null;
OutputDenseHex(offset, length, labelStr, commentStr);
}
break;
case FormatDescriptor.Type.StringGeneric:
case FormatDescriptor.Type.StringReverse:
case FormatDescriptor.Type.StringNullTerm:
case FormatDescriptor.Type.StringL8:
case FormatDescriptor.Type.StringL16:
case FormatDescriptor.Type.StringDci:
multiLine = true;
opcodeStr = operandStr = null;
OutputString(offset, labelStr, commentStr);
break;
default:
opcodeStr = "???";
operandStr = "***";
break;
}
if (!multiLine) {
opcodeStr = formatter.FormatPseudoOp(opcodeStr);
OutputLine(labelStr, opcodeStr, operandStr, commentStr);
}
}
private void OutputDenseHex(int offset, int length, string labelStr, string commentStr) {
Formatter formatter = SourceFormatter;
byte[] data = Project.FileData;
int maxPerLine = formatter.OperandWrapLen / formatter.CharsPerDenseByte;
string opcodeStr = formatter.FormatPseudoOp(sDataOpNames.Dense);
for (int i = 0; i < length; i += maxPerLine) {
int subLen = length - i;
if (subLen > maxPerLine) {
subLen = maxPerLine;
}
string operandStr = formatter.FormatDenseHex(data, offset + i, subLen);
OutputLine(labelStr, opcodeStr, operandStr, commentStr);
labelStr = commentStr = string.Empty;
}
}
/// <summary>
/// Outputs formatted data in an unformatted way, because the code generator couldn't
/// figure out how to do something better.
/// </summary>
private void OutputNoJoy(int offset, int length, string labelStr, string commentStr) {
byte[] data = Project.FileData;
Debug.Assert(length > 0);
Debug.Assert(offset >= 0 && offset < data.Length);
bool singleValue = true;
byte val = data[offset];
for (int i = 1; i < length; i++) {
if (data[offset + i] != val) {
singleValue = false;
break;
}
}
if (singleValue && length > 1) {
string opcodeStr = SourceFormatter.FormatPseudoOp(sDataOpNames.Fill);
string operandStr = length + "," + SourceFormatter.FormatHexValue(val, 2);
OutputLine(labelStr, opcodeStr, operandStr, commentStr);
} else {
OutputDenseHex(offset, length, labelStr, commentStr);
}
}
// IGenerator
public string ModifyOpcode(int offset, OpDef op) {
if (op.IsUndocumented) {
return null;
}
if (Project.CpuDef.Type == CpuDef.CpuType.CpuW65C02) {
if ((op.Opcode & 0x0f) == 0x07 || (op.Opcode & 0x0f) == 0x0f) {
// BBR, BBS, RMB, SMB not supported
return null;
}
}
// The assembler works correctly if the symbol is defined as a two-digit hex
// value (e.g. "foo equ $80") but fails if it's four (e.g. "foo equ $0080"). We
// output symbols with minimal digits, but this doesn't help if the code itself
// lives on zero page. If the operand is a reference to a zero-page user label,
// we need to output the instruction as hex.
// More info: https://github.com/apple2accumulator/merlin32/issues/8
if (op == OpDef.OpPEI_StackDPInd ||
op == OpDef.OpSTY_DPIndexX ||
op == OpDef.OpSTX_DPIndexY ||
op.AddrMode == OpDef.AddressMode.DPIndLong ||
op.AddrMode == OpDef.AddressMode.DPInd ||
op.AddrMode == OpDef.AddressMode.DPIndexXInd) {
FormatDescriptor dfd = Project.GetAnattrib(offset).DataDescriptor;
if (dfd != null && dfd.HasSymbol) {
// It has a symbol. See if the symbol target is a label (auto or user).
if (Project.SymbolTable.TryGetValue(dfd.SymbolRef.Label, out Symbol sym)) {
if (sym.IsInternalLabel) {
return null;
}
}
}
}
return string.Empty;
}
// IGenerator
public FormatDescriptor ModifyInstructionOperandFormat(int offset, FormatDescriptor dfd,
int operand) {
string badChars = ",{}";
if (dfd.FormatType == FormatDescriptor.Type.NumericLE && dfd.IsStringOrCharacter &&
badChars.IndexOf((char)(operand & 0x7f)) >= 0) {
// Merlin throws an error on certain ASCII operands, e.g. LDA #','
dfd = FormatDescriptor.Create(dfd.Length,
FormatDescriptor.Type.NumericLE, FormatDescriptor.SubType.None);
}
return dfd;
}
// IGenerator
public void UpdateCharacterEncoding(FormatDescriptor dfd) { }
// IGenerator
public void GenerateShortSequence(int offset, int length, out string opcode,
out string operand) {
Debug.Assert(length >= 1 && length <= 4);
// Use a comma-separated list of individual hex bytes.
opcode = sDataOpNames.DefineData1;
StringBuilder sb = new StringBuilder(length * 4);
for (int i = 0; i < length; i++) {
if (i != 0) {
sb.Append(',');
}
sb.Append(SourceFormatter.FormatHexValue(Project.FileData[offset + i], 2));
}
operand = sb.ToString();
}
// IGenerator
public void OutputAsmConfig() {
// nothing to do (though we could emit "xc off" for 6502)
}
// IGenerator
public void OutputEquDirective(string name, string valueStr, string comment) {
OutputLine(name, SourceFormatter.FormatPseudoOp(sDataOpNames.EquDirective),
valueStr, SourceFormatter.FormatEolComment(comment));
}
// IGenerator
public void OutputLocalVariableTable(int offset, List<DefSymbol> newDefs,
LocalVariableTable allDefs) {
foreach (DefSymbol defSym in newDefs) {
string valueStr = PseudoOp.FormatNumericOperand(SourceFormatter,
Project.SymbolTable, null, defSym.DataDescriptor, defSym.Value, 1,
PseudoOp.FormatNumericOpFlags.OmitLabelPrefixSuffix);
OutputLine(SourceFormatter.FormatVariableLabel(defSym.Label),
SourceFormatter.FormatPseudoOp(sDataOpNames.VarDirective),
valueStr, SourceFormatter.FormatEolComment(defSym.Comment));
}
}
// IGenerator
ORG rework, part 6 Added support for non-addressable regions, which are useful for things like file headers stripped out by the system loader, or chunks that get loaded into non-addressable graphics RAM. Regions are specified with the "NA" address value. The code list displays the address field greyed out, starting from zero (which is kind of handy if you want to know the relative offset within the region). Putting labels in non-addressable regions doesn't make sense, but symbol resolution is complicated enough that we really only have two options: ignore the labels entirely, or allow them but warn of their presence. The problem isn't so much the label, which you could legitimately want to access from an extension script, but rather the references to them from code or data. So we keep the label and add a warning to the Messages list when we see a reference. Moved NON_ADDR constants to Address class. AddressMap now has a copy. This is awkward because Asm65 and CommonUtil don't share. Updated the asm code generators to understand NON_ADDR, and reworked the API so that Merlin and cc65 output is correct for nested regions. Address region changes are now noted in the anattribs array, which makes certain operations faster than checking the address map. It also fixes a failure to recognize mid-instruction region changes in the code analyzer. Tweaked handling of synthetic regions, which are non-addressable areas generated by the linear address map traversal to fill in any "holes". The address region editor now treats attempts to edit them as creation of a new region.
2021-10-01 01:07:21 +00:00
public void OutputArDirective(CommonUtil.AddressMap.AddressChange change) {
int nextAddress = change.Address;
if (nextAddress == Address.NON_ADDR) {
// Start non-addressable regions at zero to ensure they don't overflow bank.
nextAddress = 0;
}
if (change.IsStart) {
AddressMap.AddressRegion region = change.Region;
if (region.HasValidPreLabel || region.HasValidIsRelative) {
// Need to output the previous ORG, if one is pending.
if (mNextAddress >= 0) {
OutputLine(string.Empty,
SourceFormatter.FormatPseudoOp(sDataOpNames.ArStartDirective),
SourceFormatter.FormatHexValue(mNextAddress, 4),
string.Empty);
}
}
if (region.HasValidPreLabel) {
string labelStr = mLocalizer.ConvLabel(change.Region.PreLabel);
OutputLine(labelStr, string.Empty, string.Empty, string.Empty);
}
if (region.HasValidIsRelative) {
// Found a valid IsRelative. Switch to "relative mode" if not there already.
mIsInRelative = true;
}
if (mIsInRelative) {
// Once we see a region with IsRelative set, we output regions as we
// find them until the next Flush.
string addrStr;
if (region.HasValidIsRelative) {
Debug.Assert(nextAddress != Address.NON_ADDR &&
region.PreLabelAddress != Address.NON_ADDR);
int diff = nextAddress - region.PreLabelAddress;
string pfxStr;
if (diff >= 0) {
pfxStr = "*+";
} else {
pfxStr = "*-";
diff = -diff;
}
addrStr = pfxStr + SourceFormatter.FormatHexValue(diff, 4);
} else {
addrStr = SourceFormatter.FormatHexValue(nextAddress, 4);
}
OutputLine(string.Empty,
SourceFormatter.FormatPseudoOp(sDataOpNames.ArStartDirective),
addrStr, string.Empty);
mNextAddress = -1;
return;
}
}
ORG rework, part 6 Added support for non-addressable regions, which are useful for things like file headers stripped out by the system loader, or chunks that get loaded into non-addressable graphics RAM. Regions are specified with the "NA" address value. The code list displays the address field greyed out, starting from zero (which is kind of handy if you want to know the relative offset within the region). Putting labels in non-addressable regions doesn't make sense, but symbol resolution is complicated enough that we really only have two options: ignore the labels entirely, or allow them but warn of their presence. The problem isn't so much the label, which you could legitimately want to access from an extension script, but rather the references to them from code or data. So we keep the label and add a warning to the Messages list when we see a reference. Moved NON_ADDR constants to Address class. AddressMap now has a copy. This is awkward because Asm65 and CommonUtil don't share. Updated the asm code generators to understand NON_ADDR, and reworked the API so that Merlin and cc65 output is correct for nested regions. Address region changes are now noted in the anattribs array, which makes certain operations faster than checking the address map. It also fixes a failure to recognize mid-instruction region changes in the code analyzer. Tweaked handling of synthetic regions, which are non-addressable areas generated by the linear address map traversal to fill in any "holes". The address region editor now treats attempts to edit them as creation of a new region.
2021-10-01 01:07:21 +00:00
mNextAddress = nextAddress;
}
// IGenerator
public void FlushArDirectives() {
// Output pending directives. There will always be something to do here unless
// we were in "relative" mode.
Debug.Assert(mNextAddress >= 0 || mIsInRelative);
if (mNextAddress >= 0) {
OutputLine(string.Empty,
SourceFormatter.FormatPseudoOp(sDataOpNames.ArStartDirective),
SourceFormatter.FormatHexValue(mNextAddress, 4),
string.Empty);
}
mNextAddress = -1;
mIsInRelative = false;
}
// IGenerator
public void OutputRegWidthDirective(int offset, int prevM, int prevX, int newM, int newX) {
// prevM/prevX may be ambiguous for offset 0, but otherwise everything
// should be either 0 or 1.
Debug.Assert(newM == 0 || newM == 1);
Debug.Assert(newX == 0 || newX == 1);
if (offset == 0 && newM == 1 && newX == 1) {
// Assembler defaults to short regs, so we can skip this.
return;
}
Various improvements The PseudoOpNames class is increasingly being used in situations where mutability is undesirable. This change makes instances immutable, eliminating the Copy() method and adding a constructor that takes a Dictionary. The serialization code now operates on a Dictionary instead of the class properties, but the JSON encoding is identical, so this doesn't invalidate app settings file data. Added an equality test to PseudoOpNames. In LineListGen, don't reset the line list if the names haven't actually changed. Use a table lookup for C64 character conversions. I figure that should be faster than multiple conditionals on a modern x64 system. Fixed a 64tass generator issue where we tried to query project properties in a call that might not have a project available (specifically, getting FormatConfig values out of the generator for use in the "quick set" buttons for Display Format). Fixed a regression test harness issue where, if the assembler reported success but didn't actually generate output, an exception would be thrown that halted the tests. Increased the width of text entry fields on the Pseudo-Op tab of app settings. The previous 8-character limit wasn't wide enough to hold ACME's "!pseudopc". Also, use TrimEnd() to remove trailing spaces (leading spaces are still allowed). In the last couple of months, Win10 started stalling for a fraction of a second when executing assemblers. It doesn't do this every time; mostly it happens if it has been a while since the assembler was run. My guess is this has to do with changes to the built-in malware scanner. Whatever the case, we now change the mouse pointer to a wait cursor while updating the assembler version cache.
2019-08-17 18:14:05 +00:00
OutputLine(string.Empty, SourceFormatter.FormatPseudoOp(REG_WIDTH_DIRECTIVE),
"%" + newM + newX, string.Empty);
}
// IGenerator
public void OutputLine(string fullLine) {
mOutStream.WriteLine(fullLine);
}
// IGenerator
public void OutputLine(string label, string opcode, string operand, string comment) {
// Split long label, but not on EQU directives (confuses the assembler).
if (mLongLabelNewLine && label.Length >= mColumnWidths[0] &&
!string.Equals(opcode, sDataOpNames.EquDirective,
StringComparison.InvariantCultureIgnoreCase)) {
mOutStream.WriteLine(label);
label = string.Empty;
}
mLineBuilder.Clear();
TextUtil.AppendPaddedString(mLineBuilder, label, 0);
TextUtil.AppendPaddedString(mLineBuilder, opcode, mColumnWidths[0]);
TextUtil.AppendPaddedString(mLineBuilder, operand,
mColumnWidths[0] + mColumnWidths[1]);
TextUtil.AppendPaddedString(mLineBuilder, comment,
mColumnWidths[0] + mColumnWidths[1] + mColumnWidths[2]);
mOutStream.WriteLine(mLineBuilder.ToString());
}
private void OutputString(int offset, string labelStr, string commentStr) {
// This gets complicated.
//
// For Dci, L8String, and L16String, the entire string needs to fit in the
// operand of one line. If it can't, we need to separate the length byte/word
// or inverted character out, and just dump the rest as ASCII. Computing the
// line length requires factoring delimiter character escapes. (NOTE: contrary
// to the documentation, STR and STRL do include trailing hex characters in the
// length calculation, so it's possible to escape delimiters.)
//
// For Reverse, we can span lines, but only if we emit the lines in
// backward order. Also, Merlin doesn't allow hex to be embedded in a REV
// operation, so we can't use REV if the string contains a delimiter.
//
// For aesthetic purposes, zero-length CString, L8String, and L16String
// should be output as DFB/DW zeroes rather than an empty string -- makes
// it easier to read.
//
// NOTE: we generally assume that the input is in the correct format, e.g.
// the length byte in a StringL8 matches dfd.Length, and the high bits in DCI strings
// have the right pattern. If not, we will generate bad output. This would need
// to be scanned and corrected at a higher level.
Anattrib attr = Project.GetAnattrib(offset);
FormatDescriptor dfd = attr.DataDescriptor;
Debug.Assert(dfd != null);
Debug.Assert(dfd.IsString);
Debug.Assert(dfd.Length > 0);
// We can sort of do parts of C64 stuff, but it's probably more readable to just
// output a commented blob than something where only the capital letters are readable.
switch (dfd.FormatSubType) {
case FormatDescriptor.SubType.Ascii:
case FormatDescriptor.SubType.HighAscii:
break;
case FormatDescriptor.SubType.C64Petscii:
case FormatDescriptor.SubType.C64Screen:
default:
OutputNoJoy(offset, dfd.Length, labelStr, commentStr);
return;
}
Formatter formatter = SourceFormatter;
byte[] data = Project.FileData;
StringOpFormatter.ReverseMode revMode = StringOpFormatter.ReverseMode.Forward;
int leadingBytes = 0;
string opcodeStr;
switch (dfd.FormatType) {
case FormatDescriptor.Type.StringGeneric:
opcodeStr = sDataOpNames.StrGeneric;
break;
case FormatDescriptor.Type.StringReverse:
opcodeStr = sDataOpNames.StrReverse;
revMode = StringOpFormatter.ReverseMode.LineReverse;
break;
case FormatDescriptor.Type.StringNullTerm:
opcodeStr = sDataOpNames.StrGeneric; // no pseudo-op for this
if (dfd.Length == 1) {
// Empty string. Just output the length byte(s) or null terminator.
GenerateShortSequence(offset, 1, out string opcode, out string operand);
OutputLine(labelStr, opcode, operand, commentStr);
return;
}
break;
case FormatDescriptor.Type.StringL8:
opcodeStr = sDataOpNames.StrLen8;
leadingBytes = 1;
break;
case FormatDescriptor.Type.StringL16:
opcodeStr = sDataOpNames.StrLen16;
leadingBytes = 2;
break;
case FormatDescriptor.Type.StringDci:
opcodeStr = sDataOpNames.StrDci;
break;
default:
Debug.Assert(false);
return;
}
// Merlin 32 uses single-quote for low ASCII, double-quote for high ASCII.
CharEncoding.Convert charConv;
char delim;
if (dfd.FormatSubType == FormatDescriptor.SubType.HighAscii) {
charConv = CharEncoding.ConvertHighAscii;
delim = '"';
} else {
charConv = CharEncoding.ConvertAscii;
delim = '\'';
}
StringOpFormatter stropf = new StringOpFormatter(SourceFormatter,
new Formatter.DelimiterDef(delim),
StringOpFormatter.RawOutputStyle.DenseHex, charConv, false);
stropf.IsDciString = (dfd.FormatType == FormatDescriptor.Type.StringDci);
// Feed bytes in, skipping over the leading length bytes.
stropf.FeedBytes(data, offset + leadingBytes,
dfd.Length - leadingBytes, 0, revMode);
Debug.Assert(stropf.Lines.Count > 0);
// See if we need to do this over.
bool redo = false;
switch (dfd.FormatType) {
case FormatDescriptor.Type.StringGeneric:
case FormatDescriptor.Type.StringNullTerm:
break;
case FormatDescriptor.Type.StringReverse:
if (stropf.HasEscapedText) {
// can't include escaped characters in REV
opcodeStr = sDataOpNames.StrGeneric;
revMode = StringOpFormatter.ReverseMode.Forward;
redo = true;
}
break;
case FormatDescriptor.Type.StringL8:
if (stropf.Lines.Count != 1) {
// single-line only
opcodeStr = sDataOpNames.StrGeneric;
leadingBytes = 1;
redo = true;
}
break;
case FormatDescriptor.Type.StringL16:
if (stropf.Lines.Count != 1) {
// single-line only
opcodeStr = sDataOpNames.StrGeneric;
leadingBytes = 2;
redo = true;
}
break;
case FormatDescriptor.Type.StringDci:
if (stropf.Lines.Count != 1) {
// single-line only
opcodeStr = sDataOpNames.StrGeneric;
stropf.IsDciString = false;
redo = true;
}
break;
default:
Debug.Assert(false);
return;
}
if (redo) {
//Debug.WriteLine("REDO off=+" + offset.ToString("x6") + ": " + dfd.FormatType);
// This time, instead of skipping over leading length bytes, we include them
// explicitly.
stropf.Reset();
stropf.FeedBytes(data, offset, dfd.Length, leadingBytes, revMode);
}
opcodeStr = formatter.FormatPseudoOp(opcodeStr);
foreach (string str in stropf.Lines) {
OutputLine(labelStr, opcodeStr, str, commentStr);
labelStr = commentStr = string.Empty; // only show on first
}
}
}
#endregion IGenerator
#region IAssembler
/// <summary>
/// Cross-assembler execution interface.
/// </summary>
public class AsmMerlin32 : IAssembler {
// Paths from generator.
private List<string> mPathNames;
// Directory to make current before executing assembler.
private string mWorkDirectory;
// IAssembler
public void GetExeIdentifiers(out string humanName, out string exeName) {
humanName = "Merlin Assembler";
exeName = "Merlin32";
}
// IAssembler
public AssemblerConfig GetDefaultConfig() {
return new AssemblerConfig(string.Empty, new int[] { 9, 6, 11, 74 });
}
// IAssembler
public AssemblerVersion QueryVersion() {
AssemblerConfig config =
AssemblerConfig.GetConfig(AppSettings.Global, AssemblerInfo.Id.Merlin32);
if (config == null || string.IsNullOrEmpty(config.ExecutablePath)) {
return null;
}
ShellCommand cmd = new ShellCommand(config.ExecutablePath, string.Empty,
Directory.GetCurrentDirectory(), null);
cmd.Execute();
if (string.IsNullOrEmpty(cmd.Stdout)) {
return null;
}
// Stdout: "C:\Src\WorkBench\Merlin32.exe v 1.0, (c) Brutal Deluxe ..."
// Other platforms may not have the ".exe". Find first occurrence of " v ".
const string PREFIX = " v "; // not expecting this to appear in the path
string str = cmd.Stdout;
int start = str.IndexOf(PREFIX);
int end = (start < 0) ? -1 : str.IndexOf(',', start);
if (start < 0 || end < 0 || start + PREFIX.Length >= end) {
Debug.WriteLine("Couldn't find version in " + str);
return null;
}
start += PREFIX.Length;
string versionStr = str.Substring(start, end - start);
CommonUtil.Version version = CommonUtil.Version.Parse(versionStr);
if (!version.IsValid) {
return null;
}
return new AssemblerVersion(versionStr, version);
}
// IAssembler
public void Configure(GenerationResults results, string workDirectory) {
// Clone pathNames, in case the caller decides to modify the original.
mPathNames = CommonUtil.Container.CopyStringList(results.PathNames);
mWorkDirectory = workDirectory;
}
// IAssembler
public AssemblerResults RunAssembler(BackgroundWorker worker) {
// Reduce input file to a partial path if possible. This is really just to make
// what we display to the user a little easier to read.
string pathName = mPathNames[0];
if (pathName.StartsWith(mWorkDirectory)) {
pathName = pathName.Remove(0, mWorkDirectory.Length + 1);
} else {
// Unexpected, but shouldn't be a problem.
Debug.WriteLine("NOTE: source file is not in work directory");
}
AssemblerConfig config =
AssemblerConfig.GetConfig(AppSettings.Global, AssemblerInfo.Id.Merlin32);
if (string.IsNullOrEmpty(config.ExecutablePath)) {
Debug.WriteLine("Assembler not configured");
return null;
}
worker.ReportProgress(0, Res.Strings.PROGRESS_ASSEMBLING);
// Wrap pathname in quotes in case it has spaces.
// (Do we need to shell-escape quotes in the pathName?)
//
// Merlin 32 has no options. The second argument is the macro include file path.
ShellCommand cmd = new ShellCommand(config.ExecutablePath, ". \"" + pathName + "\"",
mWorkDirectory, null);
cmd.Execute();
// Can't really do anything with a "cancel" request.
// Output filename is the input filename without the ".S". Since the filename
// was generated by us we can be confident in the format.
string outputFile = mPathNames[0].Substring(0, mPathNames[0].Length - 2);
return new AssemblerResults(cmd.FullCommandLine, cmd.ExitCode, cmd.Stdout,
cmd.Stderr, outputFile);
}
}
#endregion IAssembler
}