/// <summary>
/// Checks and sanitizes assembly and encoding
/// </summary>
/// <param name="check"></param>
/// <param name="encoding"></param>
/// <returns></returns>
private bool Parse_CheckEncoding(string check, string encoding, CommentSubType checkSubType)
{
//interestingly, there's at least 1 broken LLVM test
//arm64-aliases.S -> ; CHECK: tlbi alle1 ; encoding: [0x9f,0x87,0x0c,0xd5
//someone forgot to add a closing "]"
if (!encoding.StartsWith("[") /* || !encoding.EndsWith("]")*/)
{
throw new InvalidDataException($"Invalid syntax for encoding => '{encoding}'");
}
if (!encoding.EndsWith("]"))
{
//encodingTypoPresent = true;
}
encoding = encoding.Substring(1, encoding.Length - 2);
if (encoding == string.Empty)
{
return(false);
}
if (encoding.Contains("0b") || encoding.Contains("'"))
{
return(false);
}
this.Check = check;
this.Encoding = encoding;
return(true);
}
private static void PrintX86UnitTest(string assembly, string encoding, CommentSubType modifier)
{
string template = @" [Test]
public void Dis_{0}() {{
var instr = Disassemble64({1});
Assert.AreEqual(""{2}"", instr.ToString());
}}" ;
string testName = ConvertTestName(assembly);
string unitTest = string.Format(template, testName, encoding, assembly);
WriteUnitTest(
testName,
Architecture.x86, unitTest
);
}
private static void PrintMipsUnitTest(string assembly, string encoding, CommentSubType modifier)
{
string template = @" [Test]
public void MipsDis_{0}() {{
AssertCode(""{1}"", 0x{2:X8});
}}" ;
string testName = ConvertTestName(assembly);
string unitTest = string.Format(template,
testName,
assembly,
ToValueBe(encoding)
);
WriteUnitTest(
testName,
Architecture.Mips, unitTest
);
}
private static void PrintArmUnitTest(string assembly, string encoding, CommentSubType modifier)
{
string template = @" [Test]
public void ArmDasm_{0}() {{
var instr = Disassemble32(0x{2:X8});
Assert.AreEqual(""{1}"", instr.ToString());
}}" ;
string testName = ConvertTestName(assembly);
string unitTest = string.Format(template,
testName,
assembly,
ToValueBe(encoding)
);
WriteUnitTest(
testName,
Architecture.Arm, unitTest
);
}
private static void PrintAArch64UnitTest(string assembly, string encoding, CommentSubType modifier)
{
StringBuilder sb = new StringBuilder();
UInt32 valueBe = ToValueBe(encoding);
UInt32 valueLe = ToValueLe(encoding);
string template = @" [Test]
// Little-endian: {0:X8}
public void AArch64Dis_0x{1:X8}() {{
Given_Instruction(0x{1:X8});
Expect_Code(""{2}"");
}}" ;
string unitTest = string.Format(template, valueLe, valueBe, assembly);
WriteUnitTest(
Convert.ToString(valueBe),
Architecture.AArch64, unitTest
);
}
private static void PrintPowerPcUnitTest(string assembly, string encoding, CommentSubType modifier)
{
string template = @" [Test]
public void PPCDis_{0}() {{
var instr = DisassembleWord(0x{2:X8});
Assert.AreEqual(""{1}"", instr.ToString());
}}" ;
string testName = string.Format("{0:X8}", ToValueBe(encoding));
switch (modifier)
{
case CommentSubType.PPCServer:
testName += "_server";
break;
case CommentSubType.PPCEmbedded:
testName += "_embedded";
break;
case CommentSubType.PPCBigEndian:
testName += "_be";
break;
case CommentSubType.PPCLittleEndian:
testName += "_le";
break;
}
string unitTest = string.Format(template,
testName,
assembly,
ToValueBe(encoding)
);
WriteUnitTest(
testName,
Architecture.PowerPc, unitTest
);
}
private static void PrintSparcUnitTest(string assembly, string encoding, CommentSubType modifier)
{
string template = @" [Test]
public void SparcDis_{0}() {{
AssertInstruction(0x{2:X8}, ""{1}"");
}}" ;
string testName = ConvertTestName(assembly);
if (modifier == CommentSubType.SparcNoCasa)
{
testName += "_no_casa";
}
string unitTest = string.Format(template,
testName,
assembly,
ToValueBe(encoding)
);
WriteUnitTest(
testName,
Architecture.Sparc, unitTest
);
}
/// <summary>
/// Detects the type of comments for a given comment line
/// </summary>
/// <param name="line"></param>
/// <param name="type"></param>
/// <returns>The remaining part of the comment, without the leading tag (e.g CHECK)</returns>
private string DetectCommment(
string line,
out CommentType type,
out CommentSubType subType
)
{
string[] parts = line.Split(':');
subType = CommentSubType.Invalid;
if (parts.Length <= 1)
{
type = CommentType.Invalid;
return(line);
}
string commentType = parts[0];
line = string.Join(":", parts.Skip(1)).Trim();
switch (commentType)
{
case "RUN":
type = CommentType.Run;
break;
case "CHECK":
type = CommentType.Check;
break;
case "CHECK-SHA":
type = CommentType.Check;
subType = CommentSubType.Sha;
break;
case "SERVER":
type = CommentType.Check;
subType = CommentSubType.PPCServer;
break;
case "EMBEDDED":
type = CommentType.Check;
subType = CommentSubType.PPCEmbedded;
break;
case "CHECK-BE":
type = CommentType.Check;
subType = CommentSubType.PPCBigEndian;
break;
case "CHECK-LE":
type = CommentType.Check;
subType = CommentSubType.PPCLittleEndian;
break;
case "INTEL":
type = CommentType.Check;
subType = CommentSubType.IntelEncoding;
break;
case "CHECK_NO_CASA":
type = CommentType.Check;
subType = CommentSubType.SparcNoCasa;
break;
case "encoding":
type = CommentType.Encoding;
break;
default:
type = CommentType.Invalid;
break;
}
return(line);
}
/// <summary>
/// Handles a comment line
/// </summary>
/// <param name="currentLine"></param>
/// <returns></returns>
private bool HandleComment(string currentLine, out CommentSubType modifier)
{
/*
* We want to have the following layout
* lines[0] -> CHECK
* lines[1] -> CHECK encoding
* So if the comment style is OneLine, we split the line
* If the comment style is TwoLines, we read and process the next line
*/
List <string> lines = new List <string>()
{
currentLine
};
modifier = CommentSubType.Invalid;
string commentData = DetectCommment(RemoveCommentToken(currentLine), out CommentType commentType, out CommentSubType commentSubType);
if (!runVerified)
{
if (commentType == CommentType.Run)
{
if (!commentData.Contains("-show-encoding"))
{
stopHandling = true;
return(false);
}
}
runVerified = true;
}
commentType = CommentType.Invalid;
string line0, line1;
LineType line1Type;
CommentType commentType0, commentType1;
CommentSubType commentSubType0, commentSubType1;
CommentSubType checkSubType = CommentSubType.Invalid;
switch (fileCommentStyle)
{
case CommentStyle.Invalid:
return(false);
case CommentStyle.OneLine:
line0 = lines[0];
string[] parts = line0.Split(new string[] { fileCommentToken }, StringSplitOptions.RemoveEmptyEntries);
if (parts.Length < 2)
{
return(false);
}
string check = parts[0].Trim();
string encoding = parts[1].Trim();
check = DetectCommment(check, out CommentType checkType, out checkSubType);
encoding = DetectCommment(encoding, out CommentType encodingType, out CommentSubType encodingSubType);
if (checkType != CommentType.Check || encodingType != CommentType.Encoding)
{
return(false);
}
lines[0] = check;
lines.Add(encoding);
break;
case CommentStyle.TwoLines:
line1 = ReadLine(out line1Type);
if (line1Type != LineType.Comment)
{
return(false);
}
line1 = DetectCommment(RemoveCommentToken(line1), out commentType1, out commentSubType1);
if (commentType1 != CommentType.Check)
{
return(false);
}
line1 = DetectCommment(line1, out commentType1, out commentSubType1);
if (commentType1 != CommentType.Encoding)
{
return(false);
}
line0 = lines[0];
line0 = DetectCommment(RemoveCommentToken(line0), out commentType0, out commentSubType0);
switch (commentType0)
{
case CommentType.Check:
break;
default:
throw new InvalidDataException("This was supposed to be a CHECK comment");
}
checkSubType = commentSubType0;
lines[0] = line0;
lines.Add(line1);
break;
case CommentStyle.TwoLinesAsm:
line1 = ReadLine(out line1Type);
if (line1Type != LineType.Assembly)
{
return(false);
}
line0 = lines[0];
lines[0] = DetectCommment(RemoveCommentToken(line0), out commentType0, out commentSubType0);
if (commentType0 != CommentType.Check)
{
return(false);
}
checkSubType = commentSubType0;
lines.Add(line1);
// invert
string tmp = lines[0];
lines[0] = lines[1];
lines[1] = tmp;
lines[1] = DetectCommment(lines[1], out commentType1, out commentSubType1);
if (commentType1 != CommentType.Encoding)
{
return(false);
}
break;
}
/*
* Console.WriteLine($"Check => {lines[0]}");
* Console.WriteLine($"Encoding => {lines[1]}");
*/
modifier = checkSubType;
return(Parse_CheckEncoding(lines[0], lines[1], checkSubType));
}
