[MC] Support symbolic expressions in assembly directives

[phosek]

Summary

This matches the behavior of GNU assembler which supports symbolic
expressions in absolute expressions used in assembly directives.

[phosek]

This is a reland of D20656, I believe the test failure was caused by the truncateToSize invocation at MCAsmStreamer.cpp:844. The second argument is number of bytes and not number of bits to truncate to, so the original version which contained truncateToSize(Expr, 32) was obviously wrong.

Unfortunately, I haven't managed to replicate the test failure on my machines (both x86-64 Linux and OS X)—even the original version passed check-all locally but failed on bots—so I don't know if this was the only issue but I haven't spotted anything obviously wrong after going through the entire patch.

[• rafael]

Have you tried building with asan/ubsan?

Cheers,
Rafael

[phosek]

I've managed to replicate the failure in Release+Asserts build (I was using Debug previously) and I've fixed one other issue in the test, now all the tests should be passing (hopefully everywhere).

[• rafael]

Cool, go for it.

Unless there is a substantial change in the patch I don't think you
need a second review.

Cheers,
Rafael

[filcab]

Just noticed with gcc: Please add the missing emitFill overload to MCObjectStreamer too:
virtual void emitFill(uint64_t NumValues, int64_t Size, int64_t Expr);

include/llvm/MC/MCObjectStreamer.h

   bool EmitRelocDirective(const MCExpr &Offset, StringRef Name,
                           const MCExpr *Expr, SMLoc Loc) override;
   void EmitFill(uint64_t NumBytes, uint8_t FillValue) override;
+  void emitFill(const MCExpr &NumBytes, uint64_t FillValue,
+                SMLoc Loc = SMLoc()) override;
+  void emitFill(const MCExpr &NumValues, int64_t Size, int64_t Expr,
+                SMLoc Loc = SMLoc()) override;
+
   void FinishImpl() override;
 
   /// Emit the absolute difference between two symbols if possible.

include/llvm/MC/MCStreamer.h

   /// This implements directives such as '.space'.
   virtual void EmitFill(uint64_t NumBytes, uint8_t FillValue);
 
+  /// \brief Emit \p Size bytes worth of the value specified by \p FillValue.
+  ///
+  /// This is used to implement assembler directives such as .space or .skip.
+  ///
+  /// \param NumBytes - The number of bytes to emit.
+  /// \param FillValue - The value to use when filling bytes.
+  /// \param Loc - The location of the expression for error reporting.
+  virtual void emitFill(const MCExpr &NumBytes, uint64_t FillValue,
+                        SMLoc Loc = SMLoc());
+
+  /// \brief Emit \p NumValues copies of \p Size bytes. Each \p Size bytes is
+  /// taken from the lowest order 4 bytes of \p Expr expression.
+  ///
+  /// This is used to implement assembler directives such as .fill.
+  ///
+  /// \param NumValues - The number of copies of \p Size bytes to emit.
+  /// \param Size - The size (in bytes) of each repeated value.
+  /// \param Expr - The expression from which \p Size bytes are used.
+  virtual void emitFill(uint64_t NumValues, int64_t Size, int64_t Expr);
+  virtual void emitFill(const MCExpr &NumValues, int64_t Size, int64_t Expr,
+                        SMLoc Loc = SMLoc());
+
   /// \brief Emit NumBytes worth of zeros.
   /// This function properly handles data in virtual sections.
   void EmitZeros(uint64_t NumBytes);

lib/MC/MCAsmStreamer.cpp

 #include "llvm/Support/LEB128.h"
 #include "llvm/Support/MathExtras.h"
 #include "llvm/Support/Path.h"
+#include "llvm/Support/SourceMgr.h"
 #include <cctype>
 
 using namespace llvm;

 
   void EmitFill(uint64_t NumBytes, uint8_t FillValue) override;
 
+  void emitFill(const MCExpr &NumBytes, uint64_t FillValue,
+                SMLoc Loc = SMLoc()) override;
+
+  void emitFill(uint64_t NumValues, int64_t Size, int64_t Expr) override;
+
+  void emitFill(const MCExpr &NumValues, int64_t Size, int64_t Expr,
+                SMLoc Loc = SMLoc()) override;
+
   void EmitValueToAlignment(unsigned ByteAlignment, int64_t Value = 0,
                             unsigned ValueSize = 1,
                             unsigned MaxBytesToEmit = 0) override;

 void MCAsmStreamer::EmitFill(uint64_t NumBytes, uint8_t FillValue) {
   if (NumBytes == 0) return;
 
+  const MCExpr *E = MCConstantExpr::create(NumBytes, getContext());
+  emitFill(*E, FillValue);
+}
+
+void MCAsmStreamer::emitFill(const MCExpr &NumBytes, uint64_t FillValue,
+                             SMLoc Loc) {
   if (const char *ZeroDirective = MAI->getZeroDirective()) {
-    OS << ZeroDirective << NumBytes;
+    // FIXME: Emit location directives
+    OS << ZeroDirective;
+    NumBytes.print(OS, MAI);
     if (FillValue != 0)
       OS << ',' << (int)FillValue;
     EmitEOL();
     return;
   }
 
-  // Emit a byte at a time.
-  MCStreamer::EmitFill(NumBytes, FillValue);
+  MCStreamer::emitFill(NumBytes, FillValue);
+}
+
+void MCAsmStreamer::emitFill(uint64_t NumValues, int64_t Size, int64_t Expr) {
+  if (NumValues == 0)
+    return;
+
+  const MCExpr *E = MCConstantExpr::create(NumValues, getContext());
+  emitFill(*E, Size, Expr);
+}
+
+void MCAsmStreamer::emitFill(const MCExpr &NumValues, int64_t Size,
+                             int64_t Expr, SMLoc Loc) {
+  // FIXME: Emit location directives
+  OS << "\t.fill\t";
+  NumValues.print(OS, MAI);
+  OS << ", " << Size << ", 0x";
+  OS.write_hex(truncateToSize(Expr, 4));
+  EmitEOL();
 }
 
 void MCAsmStreamer::EmitValueToAlignment(unsigned ByteAlignment, int64_t Value,

lib/MC/MCObjectStreamer.cpp

 #include "llvm/MC/MCSection.h"
 #include "llvm/MC/MCSymbol.h"
 #include "llvm/Support/ErrorHandling.h"
+#include "llvm/Support/SourceMgr.h"
 #include "llvm/Support/TargetRegistry.h"
 using namespace llvm;
 

   insert(new MCFillFragment(FillValue, NumBytes));
 }
 
+void MCObjectStreamer::emitFill(const MCExpr &NumBytes, uint64_t FillValue,
+                                SMLoc Loc) {
+  MCDataFragment *DF = getOrCreateDataFragment();
+  flushPendingLabels(DF, DF->getContents().size());
+
+  int64_t IntNumBytes;
+  if (!NumBytes.evaluateAsAbsolute(IntNumBytes, getAssembler())) {
+    getContext().reportError(Loc, "expected absolute expression");
+    return;
+  }
+
+  if (IntNumBytes <= 0) {
+    getContext().reportError(Loc, "invalid number of bytes");
+    return;
+  }
+
+  EmitFill(IntNumBytes, FillValue);
+}
+
+void MCObjectStreamer::emitFill(const MCExpr &NumValues, int64_t Size,
+                                int64_t Expr, SMLoc Loc) {
+  int64_t IntNumValues;
+  if (!NumValues.evaluateAsAbsolute(IntNumValues, getAssembler())) {
+    getContext().reportError(Loc, "expected absolute expression");
+    return;
+  }
+
+  if (IntNumValues < 0) {
+    getContext().getSourceManager()->PrintMessage(
+        Loc, SourceMgr::DK_Warning,
+        "'.fill' directive with negative repeat count has no effect");
+    return;
+  }
+
+  MCStreamer::emitFill(IntNumValues, Size, Expr);
+}
+
 void MCObjectStreamer::FinishImpl() {
   // If we are generating dwarf for assembly source files dump out the sections.
   if (getContext().getGenDwarfForAssembly())

lib/MC/MCParser/AsmParser.cpp

 bool AsmParser::parseDirectiveZero() {
   checkForValidSection();
 
-  int64_t NumBytes;
-  if (parseAbsoluteExpression(NumBytes))
+  SMLoc NumBytesLoc = Lexer.getLoc();
+  const MCExpr *NumBytes;
+  if (parseExpression(NumBytes))
     return true;
 
   int64_t Val = 0;

 
   Lex();
 
-  getStreamer().EmitFill(NumBytes, Val);
+  getStreamer().emitFill(*NumBytes, Val, NumBytesLoc);
 
   return false;
 }

 bool AsmParser::parseDirectiveFill() {
   checkForValidSection();
 
-  SMLoc RepeatLoc = getLexer().getLoc();
-  int64_t NumValues;
-  if (parseAbsoluteExpression(NumValues))
+  SMLoc NumValuesLoc = Lexer.getLoc();
+  const MCExpr *NumValues;
+  if (parseExpression(NumValues))
     return true;
 
-  if (NumValues < 0) {
-    Warning(RepeatLoc,
-            "'.fill' directive with negative repeat count has no effect");
-    NumValues = 0;
-  }
-
   int64_t FillSize = 1;
   int64_t FillExpr = 0;
 

 
   if (FillSize < 0) {
     Warning(SizeLoc, "'.fill' directive with negative size has no effect");
-    NumValues = 0;
+    NumValues = MCConstantExpr::create(0, getStreamer().getContext());
   }
   if (FillSize > 8) {
     Warning(SizeLoc, "'.fill' directive with size greater than 8 has been truncated to 8");

   if (!isUInt<32>(FillExpr) && FillSize > 4)
     Warning(ExprLoc, "'.fill' directive pattern has been truncated to 32-bits");
 
-  if (NumValues > 0) {
-    int64_t NonZeroFillSize = FillSize > 4 ? 4 : FillSize;
-    FillExpr &= ~0ULL >> (64 - NonZeroFillSize * 8);
-    for (uint64_t i = 0, e = NumValues; i != e; ++i) {
-      getStreamer().EmitIntValue(FillExpr, NonZeroFillSize);
-      if (NonZeroFillSize < FillSize)
-        getStreamer().EmitIntValue(0, FillSize - NonZeroFillSize);
-    }
-  }
+  getStreamer().emitFill(*NumValues, FillSize, FillExpr, NumValuesLoc);
 
   return false;
 }

 bool AsmParser::parseDirectiveSpace(StringRef IDVal) {
   checkForValidSection();
 
-  int64_t NumBytes;
-  if (parseAbsoluteExpression(NumBytes))
+  SMLoc NumBytesLoc = Lexer.getLoc();
+  const MCExpr *NumBytes;
+  if (parseExpression(NumBytes))
     return true;
 
   int64_t FillExpr = 0;

 
   Lex();
 
-  if (NumBytes <= 0)
-    return TokError("invalid number of bytes in '" + Twine(IDVal) +
-                    "' directive");
-
   // FIXME: Sometimes the fill expr is 'nop' if it isn't supplied, instead of 0.
-  getStreamer().EmitFill(NumBytes, FillExpr);
+  getStreamer().emitFill(*NumBytes, FillExpr, NumBytesLoc);
 
   return false;
 }

lib/MC/MCStreamer.cpp

 /// EmitFill - Emit NumBytes bytes worth of the value specified by
 /// FillValue.  This implements directives such as '.space'.
 void MCStreamer::EmitFill(uint64_t NumBytes, uint8_t FillValue) {
-  const MCExpr *E = MCConstantExpr::create(FillValue, getContext());
   for (uint64_t i = 0, e = NumBytes; i != e; ++i)
-    EmitValue(E, 1);
+    EmitIntValue(FillValue, 1);
+}
+
+void MCStreamer::emitFill(uint64_t NumValues, int64_t Size, int64_t Expr) {
+  int64_t NonZeroSize = Size > 4 ? 4 : Size;
+  Expr &= ~0ULL >> (64 - NonZeroSize * 8);
+  for (uint64_t i = 0, e = NumValues; i != e; ++i) {
+    EmitIntValue(Expr, NonZeroSize);
+    if (NonZeroSize < Size)
+      EmitIntValue(0, Size - NonZeroSize);
+  }
 }
 
 /// The implementation in this class just redirects to EmitFill.

 }
 void MCStreamer::EmitULEB128Value(const MCExpr *Value) {}
 void MCStreamer::EmitSLEB128Value(const MCExpr *Value) {}
+void MCStreamer::emitFill(const MCExpr &NumBytes, uint64_t Value, SMLoc Loc) {}
+void MCStreamer::emitFill(const MCExpr &NumValues, int64_t Size, int64_t Expr,
+                          SMLoc Loc) {}
 void MCStreamer::EmitValueToAlignment(unsigned ByteAlignment, int64_t Value,
                                       unsigned ValueSize,
                                       unsigned MaxBytesToEmit) {}

test/MC/AsmParser/directive_fill.s

 # RUN: llvm-mc -triple i386-unknown-unknown %s 2> %t.err | FileCheck %s
 # RUN: FileCheck --check-prefix=CHECK-WARNINGS %s < %t.err
+# RUN: llvm-mc -triple i386-unknown-unknown -filetype=obj -o %t.o %s 2> %t.err
+# RUN: FileCheck --check-prefix=OBJ-WARNINGS %s < %t.err
 
 # CHECK: TEST0:
-# CHECK: .byte 10
+# CHECK: .fill 1, 1, 0xa
 TEST0:  
         .fill 1, 1, 10
 
 # CHECK: TEST1:
-# CHECK: .short 3
-# CHECK: .short 3
+# CHECK: .fill 2, 2, 0x3
 TEST1:  
         .fill 2, 2, 3
 
 # CHECK: TEST2:
-# CHECK: .long 4
-# CHECK: .long 0
+# CHECK: .fill 1, 8, 0x4
 TEST2:  
         .fill 1, 8, 4
 
 # CHECK: TEST3
-# CHECK: .byte 0
-# CHECK: .byte 0
-# CHECK: .byte 0
-# CHECK: .byte 0
+# CHECK: .fill 4
 TEST3:
   .fill 4
 
 # CHECK: TEST4
-# CHECK: .short 0
-# CHECK: .short 0
-# CHECK: .short 0
-# CHECK: .short 0
+# CHECK: .fill 4, 2
 TEST4:
   .fill 4, 2
 
 # CHECK: TEST5
-# CHECK: .short  2
-# CHECK: .byte   0
-# CHECK: .short  2
-# CHECK: .byte   0
-# CHECK: .short  2
-# CHECK: .byte   0
-# CHECK: .short  2
-# CHECK: .byte   0
+# CHECK: .fill 4, 3, 0x2
 TEST5:
   .fill 4, 3, 2
 
 # CHECK: TEST6
-# CHECK: .long 2
-# CHECK: .long 0
+# CHECK: .fill 1, 8, 0x2
 # CHECK-WARNINGS: '.fill' directive with size greater than 8 has been truncated to 8
 TEST6:
   .fill 1, 9, 2
 
 # CHECK: TEST7
-# CHECK: .long 0
-# CHECK: .long 0
+# CHECK: .fill 1, 8, 0x100000000
 # CHECK-WARNINGS: '.fill' directive pattern has been truncated to 32-bits
 TEST7:
   .fill 1, 8, 1<<32
 
-# CHECK-WARNINGS: '.fill' directive with negative repeat count has no effect
+# CHECK: TEST8
+# CHECK: .fill -1, 8, 0x1
+# OBJ-WARNINGS: '.fill' directive with negative repeat count has no effect
 TEST8:
   .fill -1, 8, 1
 

   .fill 1, -1, 1
 
 # CHECK: TEST10
-# CHECK: .short  22136
-# CHECK: .byte   52
+# CHECK: .fill 1, 3, 0x12345678
 TEST10:
   .fill 1, 3, 0x12345678
+
+# CHECK: TEST11
+# CHECK: .fill TEST11-TEST10, 1, 0x0
+TEST11:
+  .fill TEST11 - TEST10
+
+# CHECK: TEST12
+# CHECK: .fill TEST11-TEST12, 3, 0x12345678
+# OBJ-WARNINGS: '.fill' directive with negative repeat count has no effect
+TEST12:
+  .fill TEST11 - TEST12, 3, 0x12345678

test/MC/AsmParser/directive_space.s

 # CHECK: .space 1
 TEST2:
         .skip 1
+
+# CHECK: TEST3
+# CHECK: .space TEST0-TEST1
+TEST3:
+        .space TEST0 - TEST1