Re: [llvm-commits] [llvm-gcc] SETCC Removal Patch (Please Commit)

[Rafael Espíndola]

I don't have a normative document at hand, but I believe that in C !=
should return true if one of the operands is unordered. Currently,
llvm-gcc produces a "fcmp one".

I have patched llvm-conver.cpp:TreeToLLVM::Emit to emit a "fcmp une"
in the NE_EXPR case. The  generated code now is:

%tmp = fcmp une float %a, %b
%tmp2 = tail call bool %llvm.isunordered.f32( float %a, float %b )
%tmp3 = or bool %tmp2, %tmp

Correct, but redundant. The call to llvm.isunordered is added by
TreeToLLVM::EmitCompare. Simply removing the "if" produces the correct
code. It looks like the call to isunordered is a leftover of the old
compare instruction.

An updated patch is attached.

Comments?

Best Regards,
Rafael
___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

Re: [llvm-commits] [llvm-gcc] SETCC Removal Patch (Please Commit)

[Reid Spencer]

Rafael,

Nothing was attached to your message.

Reid.

On Sat, 2006-12-30 at 11:38 -0200, Rafael Espíndola wrote:
> I don't have a normative document at hand, but I believe that in C !=
> should return true if one of the operands is unordered. Currently,
> llvm-gcc produces a "fcmp one".
> 
> I have patched llvm-conver.cpp:TreeToLLVM::Emit to emit a "fcmp une"
> in the NE_EXPR case. The  generated code now is:
> 
> %tmp = fcmp une float %a, %b
> %tmp2 = tail call bool %llvm.isunordered.f32( float %a, float %b )
> %tmp3 = or bool %tmp2, %tmp
> 
> Correct, but redundant. The call to llvm.isunordered is added by
> TreeToLLVM::EmitCompare. Simply removing the "if" produces the correct
> code. It looks like the call to isunordered is a leftover of the old
> compare instruction.
> 
> An updated patch is attached.
> 
> Comments?
> 
> Best Regards,
> Rafael

___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

[llvm-commits] CVS: llvm/Makefile.rules

[Reid Spencer]

Changes in directory llvm:

Makefile.rules updated: 1.416 -> 1.417
---
Log message:

When compiling a C or C++ file to assembly, make the assembly output
depend on the compiler. This works around problems in the Stacker runtime
when the CFE changes in such a way that the assembly file needs to be
updated.


---
Diffs of the changes:  (+6 -6)

 Makefile.rules |   12 ++--
 1 files changed, 6 insertions(+), 6 deletions(-)


Index: llvm/Makefile.rules
diff -u llvm/Makefile.rules:1.416 llvm/Makefile.rules:1.417
--- llvm/Makefile.rules:1.416   Sat Dec 16 16:07:52 2006
+++ llvm/Makefile.rules Sat Dec 30 10:31:02 2006
@@ -1042,7 +1042,7 @@
 # Create .bc files in the ObjDir directory from .cpp .cc and .c files...
 #-
 
-$(ObjDir)/%.ll: %.cpp $(ObjDir)/.dir $(BUILT_SOURCES)
+$(ObjDir)/%.ll: %.cpp $(ObjDir)/.dir $(BUILT_SOURCES) $(LLVMGXX)
$(Echo) "Compiling $*.cpp for $(BuildMode) build (bytecode)"
$(Verb) if $(BCCompile.CXX) -MD -MT $@ -MP -MF "$(ObjDir)/$*.BCCXXd" \
   $< -o $@ -S -emit-llvm ; \
@@ -1051,7 +1051,7 @@
$(call UPGRADE_MSG,$@) 
$(call UPGRADE_LL,$@)
 
-$(ObjDir)/%.ll: %.cc $(ObjDir)/.dir $(BUILT_SOURCES)
+$(ObjDir)/%.ll: %.cc $(ObjDir)/.dir $(BUILT_SOURCES) $(LLVMGXX)
$(Echo) "Compiling $*.cc for $(BuildMode) build (bytecode)"
$(Verb) if $(BCCompile.CXX) -MD -MT $@ -MP -MF "$(ObjDir)/$*.BCCXXd" \
   $< -o $@ -S -emit-llvm ; \
@@ -1060,7 +1060,7 @@
$(call UPGRADE_MSG,$@) 
$(call UPGRADE_LL,$@)
 
-$(ObjDir)/%.ll: %.c $(ObjDir)/.dir $(BUILT_SOURCES)
+$(ObjDir)/%.ll: %.c $(ObjDir)/.dir $(BUILT_SOURCES) $(LLVMGCC)
$(Echo) "Compiling $*.c for $(BuildMode) build (bytecode)"
$(Verb) if $(BCCompile.C) -MD -MT $@ -MP -MF "$(ObjDir)/$*.BCCd" \
 $< -o $@ -S -emit-llvm ; \
@@ -1084,19 +1084,19 @@
$(Echo) "Compiling $*.c for $(BuildMode) build" $(PIC_FLAG)
$(MAYBE_PIC_Compile.C) $< -o $@ 
 
-$(ObjDir)/%.ll: %.cpp $(ObjDir)/.dir $(BUILT_SOURCES)
+$(ObjDir)/%.ll: %.cpp $(ObjDir)/.dir $(BUILT_SOURCES) $(LLVMGXX)
$(Echo) "Compiling $*.cpp for $(BuildMode) build (bytecode)"
$(BCCompile.CXX) $< -o $@ -S -emit-llvm
$(call UPGRADE_MSG,$@) 
$(call UPGRADE_LL,$@)
 
-$(ObjDir)/%.ll: %.cc $(ObjDir)/.dir $(BUILT_SOURCES)
+$(ObjDir)/%.ll: %.cc $(ObjDir)/.dir $(BUILT_SOURCES) $(LLVMGXX)
$(Echo) "Compiling $*.cc for $(BuildMode) build (bytecode)"
$(BCCompile.CXX) $< -o $@ -S -emit-llvm
$(call UPGRADE_MSG,$@) 
$(call UPGRADE_LL,$@)
 
-$(ObjDir)/%.ll: %.c $(ObjDir)/.dir $(BUILT_SOURCES)
+$(ObjDir)/%.ll: %.c $(ObjDir)/.dir $(BUILT_SOURCES) $(LLVMGCC)
$(Echo) "Compiling $*.c for $(BuildMode) build (bytecode)"
$(BCCompile.C) $< -o $@ -S -emit-llvm
$(call UPGRADE_MSG,@) 



___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

Re: [llvm-commits] [llvm-gcc] SETCC Removal Patch (Please Commit)

[Rafael Espíndola]

On 12/30/06, Reid Spencer <[EMAIL PROTECTED]> wrote:

Rafael,

Nothing was attached to your message.

oops!

Rafael
Index: gcc/llvm-convert.cpp
===
--- gcc/llvm-convert.cpp	(revisão 236)
+++ gcc/llvm-convert.cpp	(cópia de trabalho)
@@ -544,23 +544,37 @@
   case TRUTH_NOT_EXPR: Result = EmitTRUTH_NOT_EXPR(exp); break;
 
   // Binary Operators
-  case LT_EXPR:Result = EmitCompare(exp, Instruction::SetLT, 0); break;
-  case LE_EXPR:Result = EmitCompare(exp, Instruction::SetLE, 0); break;
-  case GT_EXPR:Result = EmitCompare(exp, Instruction::SetGT, 0); break;
-  case GE_EXPR:Result = EmitCompare(exp, Instruction::SetGE, 0); break;
-  case EQ_EXPR:Result = EmitCompare(exp, Instruction::SetEQ, 0); break;
-  case NE_EXPR:Result = EmitCompare(exp, Instruction::SetNE, 0); break;
-  case UNORDERED_EXPR: Result = EmitCompare(exp, 0, 1); break; // not a typo
-  case ORDERED_EXPR:   Result = EmitCompare(exp, 0, 1); break; // not a typo
-  case UNLT_EXPR:  Result = EmitCompare(exp, Instruction::SetLT, 1); break;
-  case UNLE_EXPR:  Result = EmitCompare(exp, Instruction::SetLE, 1); break;
-  case UNGT_EXPR:  Result = EmitCompare(exp, Instruction::SetGT, 1); break;
-  case UNGE_EXPR:  Result = EmitCompare(exp, Instruction::SetGE, 1); break;
-  case UNEQ_EXPR:  Result = EmitCompare(exp, Instruction::SetEQ, 1); break;
-  case LTGT_EXPR:  Result = EmitCompare(exp, Instruction::SetNE, 1); break;
-  case PLUS_EXPR:  Result = EmitBinOp(exp, DestLoc, Instruction::Add);break;
-  case MINUS_EXPR: Result = EmitBinOp(exp, DestLoc, Instruction::Sub);break;
-  case MULT_EXPR:  Result = EmitBinOp(exp, DestLoc, Instruction::Mul);break;
+  case LT_EXPR: 
+Result = EmitCompare(exp, ICmpInst::ICMP_ULT, ICmpInst::ICMP_SLT, 
+ FCmpInst::FCMP_OLT); break;
+  case LE_EXPR:
+Result = EmitCompare(exp, ICmpInst::ICMP_ULE, ICmpInst::ICMP_SLE,
+ FCmpInst::FCMP_OLE); break;
+  case GT_EXPR:
+Result = EmitCompare(exp, ICmpInst::ICMP_UGT, ICmpInst::ICMP_SGT,
+ FCmpInst::FCMP_OGT); break;
+  case GE_EXPR:
+Result = EmitCompare(exp, ICmpInst::ICMP_UGE, ICmpInst::ICMP_SGE, 
+ FCmpInst::FCMP_OGE); break;
+  case EQ_EXPR:
+Result = EmitCompare(exp, ICmpInst::ICMP_EQ, ICmpInst::ICMP_EQ, 
+ FCmpInst::FCMP_OEQ); break;
+  case NE_EXPR:
+Result = EmitCompare(exp, ICmpInst::ICMP_NE, ICmpInst::ICMP_NE, 
+ FCmpInst::FCMP_UNE); break;
+  case UNORDERED_EXPR: 
+Result = EmitCompare(exp, 0, 0, FCmpInst::FCMP_UNO); break;
+  case ORDERED_EXPR: 
+Result = EmitCompare(exp, 0, 0, FCmpInst::FCMP_ORD); break;
+  case UNLT_EXPR: Result = EmitCompare(exp, 0, 0, FCmpInst::FCMP_ULT); break;
+  case UNLE_EXPR: Result = EmitCompare(exp, 0, 0, FCmpInst::FCMP_ULE); break;
+  case UNGT_EXPR: Result = EmitCompare(exp, 0, 0, FCmpInst::FCMP_UGT); break;
+  case UNGE_EXPR: Result = EmitCompare(exp, 0, 0, FCmpInst::FCMP_UGE); break;
+  case UNEQ_EXPR: Result = EmitCompare(exp, 0, 0, FCmpInst::FCMP_UEQ); break;
+  case LTGT_EXPR: Result = EmitCompare(exp, 0, 0, FCmpInst::FCMP_UNE); break;
+  case PLUS_EXPR: Result = EmitBinOp(exp, DestLoc, Instruction::Add);break;
+  case MINUS_EXPR:Result = EmitBinOp(exp, DestLoc, Instruction::Sub);break;
+  case MULT_EXPR: Result = EmitBinOp(exp, DestLoc, Instruction::Mul);break;
   case TRUNC_DIV_EXPR: 
 if (TYPE_UNSIGNED(TREE_TYPE(exp)))
   Result = EmitBinOp(exp, DestLoc, Instruction::UDiv);
@@ -596,8 +610,14 @@
   case LROTATE_EXPR:
 Result = EmitRotateOp(exp, Instruction::Shl, Instruction::LShr);
 break;
-  case MIN_EXPR:   Result = EmitMinMaxExpr(exp, Instruction::SetLE); break;
-  case MAX_EXPR:   Result = EmitMinMaxExpr(exp, Instruction::SetGE); break;
+  case MIN_EXPR: 
+Result = EmitMinMaxExpr(exp, ICmpInst::ICMP_ULE, ICmpInst::ICMP_SLE, 
+FCmpInst::FCMP_OLE);
+break;
+  case MAX_EXPR:
+Result = EmitMinMaxExpr(exp, ICmpInst::ICMP_UGE, ICmpInst::ICMP_SGE,
+FCmpInst::FCMP_OGE);
+break;
   case CONSTRUCTOR:Result = EmitCONSTRUCTOR(exp, DestLoc); break;
 
   // Complex Math Expressions.
@@ -1353,8 +1373,8 @@
   Value *Cond = Emit(COND_EXPR_COND(exp), 0);
   // If its not already a bool, insert a comparison against zero to make it so.
   if (Cond->getType() != Type::BoolTy)
-Cond = new SetCondInst(Instruction::SetNE, Cond, 
-   Constant::getNullValue(Cond->getType()), "toBool", 
+Cond = new ICmpInst(ICmpInst::ICMP_NE, Cond, 
+Constant::getNullValue(Cond->getType()), "toBool", 
CurBB);
   tree Then = COND_EXPR_THEN(exp);
   tree Else = COND_EXPR_ELSE(exp);
@@ -2323,8 +2343,9 @@
   Value *Op = Emit(TREE_OPERAND(exp, 0), 0);
   if (!Op->getTyp

[llvm-commits] CVS: llvm/docs/Stacker.html

[Nick Lewycky]

Changes in directory llvm/docs:

Stacker.html updated: 1.22 -> 1.23
---
Log message:

Typo.


---
Diffs of the changes:  (+2 -2)

 Stacker.html |4 ++--
 1 files changed, 2 insertions(+), 2 deletions(-)


Index: llvm/docs/Stacker.html
diff -u llvm/docs/Stacker.html:1.22 llvm/docs/Stacker.html:1.23
--- llvm/docs/Stacker.html:1.22 Tue Dec 19 13:47:54 2006
+++ llvm/docs/Stacker.html  Sat Dec 30 21:44:08 2006
@@ -198,7 +198,7 @@
 Create your blocks early. While writing your compiler, you 
 will encounter several situations where you know apriori that you will
 need several blocks. For example, if-then-else, switch, while, and for
-statements in C/C++ all need multiple blocks for expression in LVVM. 
+statements in C/C++ all need multiple blocks for expression in LLVM. 
 The rule is, create them early.
 Terminate your blocks early. This just reduces the chances 
 that you forget to terminate your blocks which is required (go 
@@ -1405,7 +1405,7 @@
 
   mailto:[EMAIL PROTECTED]">Reid Spencer
   http://llvm.org";>LLVM Compiler Infrastructure
-  Last modified: $Date: 2006/12/19 19:47:54 $
+  Last modified: $Date: 2006/12/31 03:44:08 $
 
 
 



___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

[llvm-commits] CVS: llvm/include/llvm/Type.h

[Reid Spencer]

Changes in directory llvm/include/llvm:

Type.h updated: 1.96 -> 1.97
---
Log message:

For PR950: http://llvm.org/PR950 :
Make integer types signless:
  Rename [US]Byte->Int8, [U]Short->Int16, [U]Int->Int32, [U]Long->Int64
  Remove methods pertaining to sign of integer types.
  


---
Diffs of the changes:  (+11 -42)

 Type.h |   53 +++--
 1 files changed, 11 insertions(+), 42 deletions(-)


Index: llvm/include/llvm/Type.h
diff -u llvm/include/llvm/Type.h:1.96 llvm/include/llvm/Type.h:1.97
--- llvm/include/llvm/Type.h:1.96   Sat Dec 16 23:15:12 2006
+++ llvm/include/llvm/Type.hSat Dec 30 23:20:51 2006
@@ -72,12 +72,12 @@
   enum TypeID {
 // PrimitiveTypes .. make sure LastPrimitiveTyID stays up to date
 VoidTyID = 0  , BoolTyID,   //  0, 1: Basics...
-UByteTyID , SByteTyID,  //  2, 3: 8 bit types...
-UShortTyID, ShortTyID,  //  4, 5: 16 bit types...
-UIntTyID  , IntTyID,//  6, 7: 32 bit types...
-ULongTyID , LongTyID,   //  8, 9: 64 bit types...
-FloatTyID , DoubleTyID, // 10,11: Floating point types...
-LabelTyID , // 12   : Labels...
+Int8TyID,   //  2   :  8 bit type...
+Int16TyID,  //  3   : 16 bit type...
+Int32TyID,  //  4   : 32 bit type...
+Int64TyID,  //  5   : 64 bit type...
+FloatTyID, DoubleTyID,  //  6, 7: Floating point types...
+LabelTyID,  //  8   : Labels...
 
 // Derived types... see DerivedTypes.h file...
 // Make sure FirstDerivedTyID stays up to date!!!
@@ -160,28 +160,9 @@
   /// getDescription - Return the string representation of the type...
   const std::string &getDescription() const;
 
-  /// isSigned - Return whether an integral numeric type is signed.  This is
-  /// true for SByteTy, ShortTy, IntTy, LongTy.  Note that this is not true for
-  /// Float and Double.
-  ///
-  bool isSigned() const {
-return ID == SByteTyID || ID == ShortTyID ||
-   ID == IntTyID || ID == LongTyID;
-  }
-
-  /// isUnsigned - Return whether a numeric type is unsigned.  This is not 
quite
-  /// the complement of isSigned... nonnumeric types return false as they do
-  /// with isSigned.  This returns true for UByteTy, UShortTy, UIntTy, and
-  /// ULongTy
-  ///
-  bool isUnsigned() const {
-return ID == UByteTyID || ID == UShortTyID ||
-   ID == UIntTyID || ID == ULongTyID;
-  }
-
   /// isInteger - Equivalent to isSigned() || isUnsigned()
   ///
-  bool isInteger() const { return ID >= UByteTyID && ID <= LongTyID; }
+  bool isInteger() const { return ID >= Int8TyID && ID <= Int64TyID; }
 
   /// isIntegral - Returns true if this is an integral type, which is either
   /// BoolTy or one of the Integer types.
@@ -246,14 +227,6 @@
   unsigned getPrimitiveSize() const;
   unsigned getPrimitiveSizeInBits() const;
 
-  /// getUnsignedVersion - If this is an integer type, return the unsigned
-  /// variant of this type.  For example int -> uint.
-  const Type *getUnsignedVersion() const;
-
-  /// getSignedVersion - If this is an integer type, return the signed variant
-  /// of this type.  For example uint -> int.
-  const Type *getSignedVersion() const;
-  
   /// getIntegralTypeMask - Return a bitmask with ones set for all of the bits
   /// that can be set by an unsigned version of this type.  This is 0xFF for
   /// sbyte/ubyte, 0x for shorts, etc.
@@ -275,10 +248,8 @@
   /// will be promoted to if passed through a variable argument
   /// function.
   const Type *getVAArgsPromotedType() const {
-if (ID == BoolTyID || ID == UByteTyID || ID == UShortTyID)
-  return Type::UIntTy;
-else if (ID == SByteTyID || ID == ShortTyID)
-  return Type::IntTy;
+if (ID == BoolTyID || ID == Int8TyID || ID == Int16TyID)
+  return Type::Int32Ty;
 else if (ID == FloatTyID)
   return Type::DoubleTy;
 else
@@ -318,10 +289,8 @@
   // These are the builtin types that are always available...
   //
   static Type *VoidTy , *BoolTy;
-  static Type *SByteTy, *UByteTy,
-  *ShortTy, *UShortTy,
-  *IntTy  , *UIntTy,
-  *LongTy , *ULongTy;
+  static Type *Int8Ty , *Int16Ty,
+  *Int32Ty, *Int64Ty;
   static Type *FloatTy, *DoubleTy;
 
   static Type* LabelTy;



___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

[llvm-commits] CVS: llvm/include/llvm/DerivedTypes.h

[Reid Spencer]

Changes in directory llvm/include/llvm:

DerivedTypes.h updated: 1.74 -> 1.75
---
Log message:

For PR950: http://llvm.org/PR950 :
Add a new feature to FunctionType, Parameter Attributes. This allows tags
such as "sext" and "zext" to be associated with a faunction's arguments
or return type. This allows signedness information to be carried forward
from the frontend to the backend for arguments and result types.


---
Diffs of the changes:  (+43 -5)

 DerivedTypes.h |   48 +++-
 1 files changed, 43 insertions(+), 5 deletions(-)


Index: llvm/include/llvm/DerivedTypes.h
diff -u llvm/include/llvm/DerivedTypes.h:1.74 
llvm/include/llvm/DerivedTypes.h:1.75
--- llvm/include/llvm/DerivedTypes.h:1.74   Thu Dec 28 22:12:03 2006
+++ llvm/include/llvm/DerivedTypes.hSat Dec 30 23:22:12 2006
@@ -75,21 +75,40 @@
 /// FunctionType - Class to represent function types
 ///
 class FunctionType : public DerivedType {
+public:
+  /// Function parameters can have attributes to indicate how they should be
+  /// treated by optimizations and code generation. This enumeration lists the
+  /// set of possible attributes.
+  /// @brief Function parameter attributes enumeration.
+  enum ParameterAttributes {
+NoAttributeSet = 0, ///< No attribute value has been set on the parameter
+ZExtAttribute  = 1, ///< The parameter should be zero extended before call
+SExtAttribute  = 2  ///< The parameter should be sign extended before call
+  };
+  typedef std::vector ParamAttrsList;
+private:
   friend class TypeMap;
   bool isVarArgs;
+  ParamAttrsList *ParamAttrs;
 
   FunctionType(const FunctionType &);   // Do not implement
   const FunctionType &operator=(const FunctionType &);  // Do not implement
   FunctionType(const Type *Result, const std::vector &Params,
-   bool IsVarArgs);
+   bool IsVarArgs, const ParamAttrsList &Attrs);
 
 public:
   /// FunctionType::get - This static method is the primary way of constructing
-  /// a FunctionType
+  /// a FunctionType. 
   ///
-  static FunctionType *get(const Type *Result,
-   const std::vector &Params,
-   bool isVarArg);
+  static FunctionType *get(
+const Type *Result, ///< The result type
+const std::vector &Params, ///< The types of the parameters
+bool isVarArg, ///< Whether this is a variable argument length function
+const ParamAttrsList & Attrs = ParamAttrsList()
+  ///< Indicates the parameter attributes to use, if any. The 0th entry
+  ///< in the list refers to the return type. Parameters are numbered
+  ///< starting at 1. 
+  );
 
   inline bool isVarArg() const { return isVarArgs; }
   inline const Type *getReturnType() const { return ContainedTys[0]; }
@@ -106,6 +125,25 @@
   ///
   unsigned getNumParams() const { return unsigned(ContainedTys.size()-1); }
 
+  /// The parameter attributes for the \p ith parameter are returned. The 0th
+  /// parameter refers to the return type of the function.
+  /// @returns The ParameterAttributes for the \p ith parameter.
+  /// @brief Get the attributes for a parameter
+  ParameterAttributes getParamAttrs(unsigned i) const;
+
+  /// @brief Determine if a parameter attribute is set
+  bool paramHasAttr(unsigned i, ParameterAttributes attr) const {
+return getParamAttrs(i) & attr;
+  }
+
+  /// @brief Return the number of parameter attributes this type has.
+  unsigned getNumAttrs() const { 
+return (ParamAttrs ?  unsigned(ParamAttrs->size()) : 0);
+  }
+
+  /// @brief Convert a ParameterAttribute into its assembly text
+  static const char * getParamAttrsText(ParameterAttributes Attr);
+
   // Implement the AbstractTypeUser interface.
   virtual void refineAbstractType(const DerivedType *OldTy, const Type *NewTy);
   virtual void typeBecameConcrete(const DerivedType *AbsTy);



___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

[llvm-commits] CVS: llvm/include/llvm/Target/TargetLowering.h

[Reid Spencer]

Changes in directory llvm/include/llvm/Target:

TargetLowering.h updated: 1.94 -> 1.95
---
Log message:

For PR950: http://llvm.org/PR950 :
Change integer type names for signless integer types


---
Diffs of the changes:  (+13 -12)

 TargetLowering.h |   25 +
 1 files changed, 13 insertions(+), 12 deletions(-)


Index: llvm/include/llvm/Target/TargetLowering.h
diff -u llvm/include/llvm/Target/TargetLowering.h:1.94 
llvm/include/llvm/Target/TargetLowering.h:1.95
--- llvm/include/llvm/Target/TargetLowering.h:1.94  Wed Dec 13 14:52:00 2006
+++ llvm/include/llvm/Target/TargetLowering.h   Sat Dec 30 23:23:18 2006
@@ -343,14 +343,10 @@
 default: assert(0 && "Unknown type!");
 case Type::VoidTyID:return MVT::isVoid;
 case Type::BoolTyID:return MVT::i1;
-case Type::UByteTyID:
-case Type::SByteTyID:   return MVT::i8;
-case Type::ShortTyID:
-case Type::UShortTyID:  return MVT::i16;
-case Type::IntTyID:
-case Type::UIntTyID:return MVT::i32;
-case Type::LongTyID:
-case Type::ULongTyID:   return MVT::i64;
+case Type::Int8TyID:return MVT::i8;
+case Type::Int16TyID:   return MVT::i16;
+case Type::Int32TyID:   return MVT::i32;
+case Type::Int64TyID:   return MVT::i64;
 case Type::FloatTyID:   return MVT::f32;
 case Type::DoubleTyID:  return MVT::f64;
 case Type::PointerTyID: return PointerTy;
@@ -743,11 +739,16 @@
   /// actual call.  This returns a pair of operands.  The first element is the
   /// return value for the function (if RetTy is not VoidTy).  The second
   /// element is the outgoing token chain.
-  typedef std::vector > ArgListTy;
+  struct ArgListEntry {
+SDOperand Node;
+const Type* Ty;
+bool isSigned;
+  };
+  typedef std::vector ArgListTy;
   virtual std::pair
-  LowerCallTo(SDOperand Chain, const Type *RetTy, bool isVarArg,
-  unsigned CallingConv, bool isTailCall, SDOperand Callee,
-  ArgListTy &Args, SelectionDAG &DAG);
+  LowerCallTo(SDOperand Chain, const Type *RetTy, bool RetTyIsSigned, 
+  bool isVarArg, unsigned CallingConv, bool isTailCall, 
+  SDOperand Callee, ArgListTy &Args, SelectionDAG &DAG);
 
   /// LowerFrameReturnAddress - This hook lowers a call to llvm.returnaddress 
or
   /// llvm.frameaddress (depending on the value of the first argument).  The



___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

[llvm-commits] CVS: llvm/include/llvm/InstrTypes.h Intrinsics.td

[Reid Spencer]

Changes in directory llvm/include/llvm:

InstrTypes.h updated: 1.57 -> 1.58
Intrinsics.td updated: 1.35 -> 1.36
---
Log message:

For PR950: http://llvm.org/PR950 :
Change integer type names for signless integer types


---
Diffs of the changes:  (+9 -9)

 InstrTypes.h  |2 +-
 Intrinsics.td |   16 
 2 files changed, 9 insertions(+), 9 deletions(-)


Index: llvm/include/llvm/InstrTypes.h
diff -u llvm/include/llvm/InstrTypes.h:1.57 llvm/include/llvm/InstrTypes.h:1.58
--- llvm/include/llvm/InstrTypes.h:1.57 Sat Dec 23 13:06:54 2006
+++ llvm/include/llvm/InstrTypes.h  Sat Dec 30 23:23:18 2006
@@ -427,7 +427,7 @@
   /// involving Integer and Pointer types. They are no-op casts if the integer
   /// is the same size as the pointer. However, pointer size varies with 
   /// platform. Generally, the result of TargetData::getIntPtrType() should be
-  /// passed in. If that's not available, use Type::ULongTy, which will make
+  /// passed in. If that's not available, use Type::Int64Ty, which will make
   /// the isNoopCast call conservative.
   /// @brief Determine if this cast is a no-op cast. 
   bool isNoopCast(


Index: llvm/include/llvm/Intrinsics.td
diff -u llvm/include/llvm/Intrinsics.td:1.35 
llvm/include/llvm/Intrinsics.td:1.36
--- llvm/include/llvm/Intrinsics.td:1.35Fri Sep  8 01:43:00 2006
+++ llvm/include/llvm/Intrinsics.td Sat Dec 30 23:23:18 2006
@@ -65,14 +65,14 @@
 
 def llvm_void_ty   : LLVMType;
 def llvm_bool_ty   : LLVMType;
-def llvm_sbyte_ty  : LLVMType;
-def llvm_short_ty  : LLVMType;
-def llvm_int_ty: LLVMType;
-def llvm_long_ty   : LLVMType;
-def llvm_ubyte_ty  : LLVMType;
-def llvm_ushort_ty : LLVMType;
-def llvm_uint_ty   : LLVMType;
-def llvm_ulong_ty  : LLVMType;
+def llvm_sbyte_ty  : LLVMType;
+def llvm_short_ty  : LLVMType;
+def llvm_int_ty: LLVMType;
+def llvm_long_ty   : LLVMType;
+def llvm_ubyte_ty  : LLVMType;
+def llvm_ushort_ty : LLVMType;
+def llvm_uint_ty   : LLVMType;
+def llvm_ulong_ty  : LLVMType;
 def llvm_float_ty  : LLVMType;
 def llvm_double_ty : LLVMType;
 def llvm_ptr_ty: LLVMType; // sbyte*



___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

[llvm-commits] CVS: llvm/include/llvm/ExecutionEngine/GenericValue.h

[Reid Spencer]

Changes in directory llvm/include/llvm/ExecutionEngine:

GenericValue.h updated: 1.7 -> 1.8
---
Log message:

For PR950: http://llvm.org/PR950 :
Change integer type names for signless integer types


---
Diffs of the changes:  (+4 -8)

 GenericValue.h |   12 
 1 files changed, 4 insertions(+), 8 deletions(-)


Index: llvm/include/llvm/ExecutionEngine/GenericValue.h
diff -u llvm/include/llvm/ExecutionEngine/GenericValue.h:1.7 
llvm/include/llvm/ExecutionEngine/GenericValue.h:1.8
--- llvm/include/llvm/ExecutionEngine/GenericValue.h:1.7Thu Apr 21 
15:39:54 2005
+++ llvm/include/llvm/ExecutionEngine/GenericValue.hSat Dec 30 23:23:18 2006
@@ -23,14 +23,10 @@
 
 union GenericValue {
   boolBoolVal;
-  unsigned char   UByteVal;
-  signed   char   SByteVal;
-  unsigned short  UShortVal;
-  signed   short  ShortVal;
-  unsigned intUIntVal;
-  signed   intIntVal;
-  uint64_tULongVal;
-  int64_t LongVal;
+  unsigned char   Int8Val;
+  unsigned short  Int16Val;
+  unsigned intInt32Val;
+  uint64_tInt64Val;
   double  DoubleVal;
   float   FloatVal;
   struct { unsigned int first; unsigned int second; } UIntPairVal;



___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

[llvm-commits] CVS: llvm/lib/VMCore/AsmWriter.cpp

[Reid Spencer]

Changes in directory llvm/lib/VMCore:

AsmWriter.cpp updated: 1.231 -> 1.232
---
Log message:

For PR950: http://llvm.org/PR950 :
* Change integer type name from signed to signless
* Implement printing of FunctionType parameter attributes.


---
Diffs of the changes:  (+49 -16)

 AsmWriter.cpp |   65 +++---
 1 files changed, 49 insertions(+), 16 deletions(-)


Index: llvm/lib/VMCore/AsmWriter.cpp
diff -u llvm/lib/VMCore/AsmWriter.cpp:1.231 llvm/lib/VMCore/AsmWriter.cpp:1.232
--- llvm/lib/VMCore/AsmWriter.cpp:1.231 Fri Dec 29 14:29:48 2006
+++ llvm/lib/VMCore/AsmWriter.cpp   Sat Dec 30 23:24:50 2006
@@ -271,12 +271,22 @@
   case Type::FunctionTyID: {
 const FunctionType *FTy = cast(Ty);
 calcTypeName(FTy->getReturnType(), TypeStack, TypeNames, Result);
+if (FTy->getParamAttrs(0)) {
+  Result += " ";
+  Result += FunctionType::getParamAttrsText(FTy->getParamAttrs(0));
+}
 Result += " (";
+unsigned Idx = 1;
 for (FunctionType::param_iterator I = FTy->param_begin(),
E = FTy->param_end(); I != E; ++I) {
   if (I != FTy->param_begin())
 Result += ", ";
   calcTypeName(*I, TypeStack, TypeNames, Result);
+  if (FTy->getParamAttrs(Idx)) {
+Result += + " ";
+Result += FunctionType::getParamAttrsText(FTy->getParamAttrs(Idx));
+  }
+  Idx++;
 }
 if (FTy->isVarArg()) {
   if (FTy->getNumParams()) Result += ", ";
@@ -660,7 +670,7 @@
   void printConstant(const Constant *CPV);
   void printGlobal(const GlobalVariable *GV);
   void printFunction(const Function *F);
-  void printArgument(const Argument *FA);
+  void printArgument(const Argument *FA, FunctionType::ParameterAttributes A);
   void printBasicBlock(const BasicBlock *BB);
   void printInstruction(const Instruction &I);
 
@@ -687,12 +697,20 @@
 ///
 std::ostream &AssemblyWriter::printTypeAtLeastOneLevel(const Type *Ty) {
   if (const FunctionType *FTy = dyn_cast(Ty)) {
-printType(FTy->getReturnType()) << " (";
+printType(FTy->getReturnType());
+if (FTy->getParamAttrs(0))
+  Out << ' ' << FunctionType::getParamAttrsText(FTy->getParamAttrs(0));
+Out << " (";
+unsigned Idx = 1;
 for (FunctionType::param_iterator I = FTy->param_begin(),
E = FTy->param_end(); I != E; ++I) {
   if (I != FTy->param_begin())
 Out << ", ";
   printType(*I);
+  if (FTy->getParamAttrs(Idx)) {
+Out << " " << FunctionType::getParamAttrsText(FTy->getParamAttrs(Idx));
+  }
+  Idx++;
 }
 if (FTy->isVarArg()) {
   if (FTy->getNumParams()) Out << ", ";
@@ -949,7 +967,10 @@
   default: Out << "cc" << F->getCallingConv() << " "; break;
   }
 
+  const FunctionType *FT = F->getFunctionType();
   printType(F->getReturnType()) << ' ';
+  if (FT->getParamAttrs(0))
+Out << FunctionType::getParamAttrsText(FT->getParamAttrs(0)) << ' ';
   if (!F->getName().empty())
 Out << getLLVMName(F->getName());
   else
@@ -958,11 +979,15 @@
   Machine.incorporateFunction(F);
 
   // Loop over the arguments, printing them...
-  const FunctionType *FT = F->getFunctionType();
 
+  unsigned Idx = 1;
   for (Function::const_arg_iterator I = F->arg_begin(), E = F->arg_end();
-   I != E; ++I)
-printArgument(I);
+   I != E; ++I) {
+// Insert commas as we go... the first arg doesn't get a comma
+if (I != F->arg_begin()) Out << ", ";
+printArgument(I, FT->getParamAttrs(Idx));
+Idx++;
+  }
 
   // Finish printing arguments...
   if (FT->isVarArg()) {
@@ -994,13 +1019,14 @@
 /// printArgument - This member is called for every argument that is passed 
into
 /// the function.  Simply print it out
 ///
-void AssemblyWriter::printArgument(const Argument *Arg) {
-  // Insert commas as we go... the first arg doesn't get a comma
-  if (Arg != Arg->getParent()->arg_begin()) Out << ", ";
-
+void AssemblyWriter::printArgument(const Argument *Arg, 
+   FunctionType::ParameterAttributes attrs) {
   // Output type...
   printType(Arg->getType());
 
+  if (attrs != FunctionType::NoAttributeSet)
+Out << ' ' << FunctionType::getParamAttrsText(attrs);
+
   // Output name, if available...
   if (Arg->hasName())
 Out << ' ' << getLLVMName(Arg->getName());
@@ -1160,17 +1186,20 @@
 (!isa(RetTy) ||
  !isa(cast(RetTy)->getElementType( {
   Out << ' '; printType(RetTy);
+  if (FTy->getParamAttrs(0) != FunctionType::NoAttributeSet)
+Out << " " << FTy->getParamAttrsText(FTy->getParamAttrs(0));
   writeOperand(Operand, false);
 } else {
   writeOperand(Operand, true);
 }
 Out << '(';
-if (CI->getNumOperands() > 1) writeOperand(CI->getOperand(1), true);
-for (unsigned op = 2, Eop = I.getNumOperands(); op < Eop; ++op) {
-  Out << ',';
+for (unsigned op = 1, Eop = I.getNumOperands(); op < Eop; ++op) {
+  if (op > 1)
+Out << ',';
   writeOperand(I.getOpe

[llvm-commits] CVS: llvm/lib/VMCore/Type.cpp

[Reid Spencer]

Changes in directory llvm/lib/VMCore:

Type.cpp updated: 1.152 -> 1.153
---
Log message:

For PR950: http://llvm.org/PR950 :
Implement signless integer types and FunctionType parameter attributes.


---
Diffs of the changes:  (+90 -98)

 Type.cpp |  188 ++-
 1 files changed, 90 insertions(+), 98 deletions(-)


Index: llvm/lib/VMCore/Type.cpp
diff -u llvm/lib/VMCore/Type.cpp:1.152 llvm/lib/VMCore/Type.cpp:1.153
--- llvm/lib/VMCore/Type.cpp:1.152  Fri Dec  8 12:06:16 2006
+++ llvm/lib/VMCore/Type.cppSat Dec 30 23:25:34 2006
@@ -74,14 +74,10 @@
   switch (IDNumber) {
   case VoidTyID  : return VoidTy;
   case BoolTyID  : return BoolTy;
-  case UByteTyID : return UByteTy;
-  case SByteTyID : return SByteTy;
-  case UShortTyID: return UShortTy;
-  case ShortTyID : return ShortTy;
-  case UIntTyID  : return UIntTy;
-  case IntTyID   : return IntTy;
-  case ULongTyID : return ULongTy;
-  case LongTyID  : return LongTy;
+  case Int8TyID  : return Int8Ty; 
+  case Int16TyID : return Int16Ty; 
+  case Int32TyID : return Int32Ty;
+  case Int64TyID : return Int64Ty;
   case FloatTyID : return FloatTy;
   case DoubleTyID: return DoubleTy;
   case LabelTyID : return LabelTy;
@@ -120,57 +116,11 @@
   // At this point we have only various mismatches of the first class types
   // remaining and ptr->ptr. Just select the lossless conversions. Everything
   // else is not lossless.
-  switch (getTypeID()) {
-  case Type::UByteTyID:   return Ty == Type::SByteTy;
-  case Type::SByteTyID:   return Ty == Type::UByteTy;
-  case Type::UShortTyID:  return Ty == Type::ShortTy;
-  case Type::ShortTyID:   return Ty == Type::UShortTy;
-  case Type::UIntTyID:return Ty == Type::IntTy;
-  case Type::IntTyID: return Ty == Type::UIntTy;
-  case Type::ULongTyID:   return Ty == Type::LongTy;
-  case Type::LongTyID:return Ty == Type::ULongTy;
-  case Type::PointerTyID: return isa(Ty);
-  default:
-break;
-  }
+  if (getTypeID() == Type::PointerTyID)
+return isa(Ty);
   return false;  // Other types have no identity values
 }
 
-/// getUnsignedVersion - If this is an integer type, return the unsigned
-/// variant of this type.  For example int -> uint.
-const Type *Type::getUnsignedVersion() const {
-  switch (getTypeID()) {
-  default:
-assert(isInteger()&&"Type::getUnsignedVersion is only valid for 
integers!");
-  case Type::UByteTyID:
-  case Type::SByteTyID:   return Type::UByteTy;
-  case Type::UShortTyID:
-  case Type::ShortTyID:   return Type::UShortTy;
-  case Type::UIntTyID:
-  case Type::IntTyID: return Type::UIntTy;
-  case Type::ULongTyID:
-  case Type::LongTyID:return Type::ULongTy;
-  }
-}
-
-/// getSignedVersion - If this is an integer type, return the signed variant
-/// of this type.  For example uint -> int.
-const Type *Type::getSignedVersion() const {
-  switch (getTypeID()) {
-  default:
-assert(isInteger() && "Type::getSignedVersion is only valid for 
integers!");
-  case Type::UByteTyID:
-  case Type::SByteTyID:   return Type::SByteTy;
-  case Type::UShortTyID:
-  case Type::ShortTyID:   return Type::ShortTy;
-  case Type::UIntTyID:
-  case Type::IntTyID: return Type::IntTy;
-  case Type::ULongTyID:
-  case Type::LongTyID:return Type::LongTy;
-  }
-}
-
-
 // getPrimitiveSize - Return the basic size of this type if it is a primitive
 // type.  These are fixed by LLVM and are not target dependent.  This will
 // return zero if the type does not have a size or is not a primitive type.
@@ -178,15 +128,11 @@
 unsigned Type::getPrimitiveSize() const {
   switch (getTypeID()) {
   case Type::BoolTyID:
-  case Type::SByteTyID:
-  case Type::UByteTyID: return 1;
-  case Type::UShortTyID:
-  case Type::ShortTyID: return 2;
+  case Type::Int8TyID:  return 1;
+  case Type::Int16TyID: return 2;
   case Type::FloatTyID:
-  case Type::IntTyID:
-  case Type::UIntTyID: return 4;
-  case Type::LongTyID:
-  case Type::ULongTyID:
+  case Type::Int32TyID: return 4;
+  case Type::Int64TyID:
   case Type::DoubleTyID: return 8;
   default: return 0;
   }
@@ -195,15 +141,11 @@
 unsigned Type::getPrimitiveSizeInBits() const {
   switch (getTypeID()) {
   case Type::BoolTyID:  return 1;
-  case Type::SByteTyID:
-  case Type::UByteTyID: return 8;
-  case Type::UShortTyID:
-  case Type::ShortTyID: return 16;
+  case Type::Int8TyID:  return 8;
+  case Type::Int16TyID: return 16;
   case Type::FloatTyID:
-  case Type::IntTyID:
-  case Type::UIntTyID: return 32;
-  case Type::LongTyID:
-  case Type::ULongTyID:
+  case Type::Int32TyID:return 32;
+  case Type::Int64TyID:
   case Type::DoubleTyID: return 64;
   case Type::PackedTyID: {
 const PackedType *PTy = cast(this);
@@ -303,11 +245,21 @@
   switch (Ty->getTypeID()) {
   case Type::FunctionTyID: {
 const FunctionType *FTy = cast(Ty);
-Result = getTypeDescription(FTy->getReturnType(), TypeStack) + " (";
+Result = FunctionType::getParamAttrsText(FTy->getParamAttrs(0));
+

[llvm-commits] CVS: llvm/lib/VMCore/ConstantFolding.cpp Constants.cpp Function.cpp Instructions.cpp Module.cpp ValueTypes.cpp Verifier.cpp

[Reid Spencer]

Changes in directory llvm/lib/VMCore:

ConstantFolding.cpp updated: 1.121 -> 1.122
Constants.cpp updated: 1.192 -> 1.193
Function.cpp updated: 1.107 -> 1.108
Instructions.cpp updated: 1.57 -> 1.58
Module.cpp updated: 1.69 -> 1.70
ValueTypes.cpp updated: 1.8 -> 1.9
Verifier.cpp updated: 1.179 -> 1.180
---
Log message:

For PR950: http://llvm.org/PR950 :
Change signed integer type names to unsigned equivalents.


---
Diffs of the changes:  (+82 -137)

 ConstantFolding.cpp |   37 +++-
 Constants.cpp   |  117 +---
 Function.cpp|2 
 Instructions.cpp|   38 +---
 Module.cpp  |   15 +++---
 ValueTypes.cpp  |8 +--
 Verifier.cpp|2 
 7 files changed, 82 insertions(+), 137 deletions(-)


Index: llvm/lib/VMCore/ConstantFolding.cpp
diff -u llvm/lib/VMCore/ConstantFolding.cpp:1.121 
llvm/lib/VMCore/ConstantFolding.cpp:1.122
--- llvm/lib/VMCore/ConstantFolding.cpp:1.121   Sun Dec 24 12:52:08 2006
+++ llvm/lib/VMCore/ConstantFolding.cpp Sat Dec 30 23:26:44 2006
@@ -87,7 +87,7 @@
   for (unsigned i = 0; i != SrcNumElts; ++i) {
 uint64_t V =
   DoubleToBits(cast(CP->getOperand(i))->getValue());
-Constant *C = ConstantInt::get(Type::ULongTy, V);
+Constant *C = ConstantInt::get(Type::Int64Ty, V);
 Result.push_back(ConstantExpr::getBitCast(C, DstEltTy ));
   }
   return ConstantPacked::get(Result);
@@ -96,7 +96,7 @@
 assert(SrcEltTy->getTypeID() == Type::FloatTyID);
 for (unsigned i = 0; i != SrcNumElts; ++i) {
   uint32_t V = 
FloatToBits(cast(CP->getOperand(i))->getValue());
-  Constant *C = ConstantInt::get(Type::UIntTy, V);
+  Constant *C = ConstantInt::get(Type::Int32Ty, V);
   Result.push_back(ConstantExpr::getBitCast(C, DstEltTy));
 }
 return ConstantPacked::get(Result);
@@ -132,7 +132,7 @@
 
   // Let CastInst::isEliminableCastPair do the heavy lifting.
   return CastInst::isEliminableCastPair(firstOp, secondOp, SrcTy, MidTy, DstTy,
-Type::ULongTy);
+Type::Int64Ty);
 }
 
 Constant *llvm::ConstantFoldCastInstruction(unsigned opc, const Constant *V,
@@ -217,13 +217,13 @@
 if (const PointerType *PTy = dyn_cast(V->getType()))
   if (const PointerType *DPTy = dyn_cast(DestTy)) {
 std::vector IdxList;
-IdxList.push_back(Constant::getNullValue(Type::IntTy));
+IdxList.push_back(Constant::getNullValue(Type::Int32Ty));
 const Type *ElTy = PTy->getElementType();
 while (ElTy != DPTy->getElementType()) {
   if (const StructType *STy = dyn_cast(ElTy)) {
 if (STy->getNumElements() == 0) break;
 ElTy = STy->getElementType(0);
-IdxList.push_back(Constant::getNullValue(Type::UIntTy));
+IdxList.push_back(Constant::getNullValue(Type::Int32Ty));
   } else if (const SequentialType *STy = 
  dyn_cast(ElTy)) {
 if (isa(ElTy)) break;  // Can't index into pointers!
@@ -296,10 +296,10 @@
 if (const ConstantFP *FP = dyn_cast(V)) {
   // FP -> Integral.
   if (DestTy->isIntegral()) {
-if (DestTy == Type::IntTy || DestTy == Type::UIntTy)
+if (DestTy == Type::Int32Ty)
   return ConstantInt::get(DestTy, FloatToBits(FP->getValue()));
-assert((DestTy == Type::LongTy || DestTy == Type::ULongTy) 
-   && "Incorrect integer  type for bitcast!");
+assert(DestTy == Type::Int64Ty && 
+   "Incorrect integer  type for bitcast!");
 return ConstantInt::get(DestTy, DoubleToBits(FP->getValue()));
   }
 }
@@ -712,16 +712,13 @@
 
   // Ok, we have two differing integer indices.  Sign extend them to be the 
same
   // type.  Long is always big enough, so we use it.
-  if (C1->getType() != Type::LongTy && C1->getType() != Type::ULongTy)
-C1 = ConstantExpr::getSExt(C1, Type::LongTy);
-  else
-C1 = ConstantExpr::getBitCast(C1, Type::LongTy);
-  if (C2->getType() != Type::LongTy && C1->getType() != Type::ULongTy)
-C2 = ConstantExpr::getSExt(C2, Type::LongTy);
-  else
-C2 = ConstantExpr::getBitCast(C2, Type::LongTy);
+  if (C1->getType() != Type::Int64Ty)
+C1 = ConstantExpr::getSExt(C1, Type::Int64Ty);
+
+  if (C2->getType() != Type::Int64Ty)
+C1 = ConstantExpr::getSExt(C2, Type::Int64Ty);
 
-  if (C1 == C2) return 0;  // Are they just differing types?
+  if (C1 == C2) return 0;  // They are equal
 
   // If the type being indexed over is really just a zero sized type, there is
   // no pointer difference being made here.
@@ -1324,7 +1321,7 @@
   if (uint32_t ElSize = ElTy->getPrimitiveSize()) {
 // gep null, C is equal to C*sizeof(nullty).  If nullty is a known llvm
 // type, we can statically fold this.
-Constant *R = ConstantInt::get(Type::UIntTy, ElSize);
+Constant *R = ConstantInt::get(Type::Int32

[llvm-commits] CVS: llvm/lib/AsmParser/Lexer.l llvmAsmParser.y

[Reid Spencer]

Changes in directory llvm/lib/AsmParser:

Lexer.l updated: 1.89 -> 1.90
llvmAsmParser.y updated: 1.292 -> 1.293
---
Log message:

For PR950: http://llvm.org/PR950 :
Major reorganization. This patch introduces the signedness changes for
the new integer types (i8, i16, i32, i64) which replace the old signed
versions (ubyte, sbyte, ushort, short, etc). This patch also implements
the function type parameter attributes feature. Together these conspired
to introduce new reduce/reduce errors into the grammar. Consequently, it
was necessary to introduce a new keyword into the grammar in order to
disambiguate. Without this, yacc would make incorrect shift/reduce and
reduce/reduce decisions and fail to parse the intended assembly.

Changes in assembly:

1. The "implementation" keyword is superfluous but still supported. You
   can use it as a sentry which will ensure there are no remaining up
   reference types. However, this is optional as those checks are also
   performed elsewhere.

2. Parameter attributes are now implemented using an at sign to
   indicate the attribute. The attributes are placed after the type
   in a function declaration or after the argument value in a function
   call. For example:
  i8 @sext %myfunc(i16 @zext)
  call i8 @sext %myfunc(i16 @zext %someVal)
   The facility is available for supporting additional attributes and
   they can be combined using the @(attr1,attr2,attr3) syntax. Right
   now  the only two supported are @sext and @zext

3. Functions must now be defined with the "define" keyword which is
   analagous to the "declare" keyword for function declarations. The
   introduction of this keyword disambiguates situations where a 
   named result type is confused with a new type or gvar definition.
   For example:
  %MyType = type i16
  %MyType %func(%MyType) { ... }
   With the introduction of optional parameter attributes between 
   the function name and the function result type, yacc will pick
   the wrong rule to reduce unless it is disambiguated with "define"
   before the function definition, as in:
  define %MyType @zext %func(%MyType %someArg) { ... }



---
Diffs of the changes:  (+413 -253)

 Lexer.l |   12 -
 llvmAsmParser.y |  654 +++-
 2 files changed, 413 insertions(+), 253 deletions(-)


Index: llvm/lib/AsmParser/Lexer.l
diff -u llvm/lib/AsmParser/Lexer.l:1.89 llvm/lib/AsmParser/Lexer.l:1.90
--- llvm/lib/AsmParser/Lexer.l:1.89 Fri Dec 29 14:29:48 2006
+++ llvm/lib/AsmParser/Lexer.l  Sat Dec 30 23:40:12 2006
@@ -235,14 +235,10 @@
 
 void{ RET_TY(Type::VoidTy,  VOID);  }
 bool{ RET_TY(Type::BoolTy,  BOOL);  }
-sbyte   { RET_TY(Type::SByteTy, SBYTE); }
-ubyte   { RET_TY(Type::UByteTy, UBYTE); }
-short   { RET_TY(Type::ShortTy, SHORT); }
-ushort  { RET_TY(Type::UShortTy,USHORT);}
-int { RET_TY(Type::IntTy,   INT);   }
-uint{ RET_TY(Type::UIntTy,  UINT);  }
-long{ RET_TY(Type::LongTy,  LONG);  }
-ulong   { RET_TY(Type::ULongTy, ULONG); }
+i8  { RET_TY(Type::Int8Ty,  INT8);  }
+i16 { RET_TY(Type::Int16Ty, INT16); }
+i32 { RET_TY(Type::Int32Ty, INT32); }
+i64 { RET_TY(Type::Int64Ty, INT64); }
 float   { RET_TY(Type::FloatTy, FLOAT); }
 double  { RET_TY(Type::DoubleTy,DOUBLE);}
 label   { RET_TY(Type::LabelTy, LABEL); }


Index: llvm/lib/AsmParser/llvmAsmParser.y
diff -u llvm/lib/AsmParser/llvmAsmParser.y:1.292 
llvm/lib/AsmParser/llvmAsmParser.y:1.293
--- llvm/lib/AsmParser/llvmAsmParser.y:1.292Fri Dec 29 14:29:48 2006
+++ llvm/lib/AsmParser/llvmAsmParser.y  Sat Dec 30 23:40:12 2006
@@ -19,12 +19,16 @@
 #include "llvm/Module.h"
 #include "llvm/SymbolTable.h"
 #include "llvm/Support/GetElementPtrTypeIterator.h"
+#include "llvm/Support/CommandLine.h"
 #include "llvm/ADT/STLExtras.h"
 #include "llvm/Support/MathExtras.h"
 #include "llvm/Support/Streams.h"
 #include 
 #include 
 #include 
+#ifndef NDEBUG
+#define YYDEBUG 1
+#endif
 
 // The following is a gross hack. In order to rid the libAsmParser library of
 // exceptions, we have to have a way of getting the yyparse function to go into
@@ -47,6 +51,11 @@
 
 namespace llvm {
   std::string CurFilename;
+#if YYDEBUG
+static cl::opt
+Debug("debug-yacc", cl::desc("Print yacc debug state changes"), 
+  cl::Hidden, cl::init(false));
+#endif
 }
 using namespace llvm;
 
@@ -71,6 +80,7 @@
 // destroyed when the function is completed.
 //
 typedef std::vector ValueList;   // Numbered defs
+
 static void 
 ResolveDefinitions(std::map &LateResolvers,
std::map *FutureLateResolvers = 0);
@@ -804,34 +814,12 @@
   return Ty;
 }
 
-// common code from the two 'RunVMAsmParser' functions
-static Module* RunParser(Module * M) {
-
-  llvmAsmlineno = 1;  // Reset the current line number...
-  CurModule.CurrentModule = M;
-
-  // Check to make sure the

[llvm-commits] CVS: llvm/lib/Bytecode/Reader/Reader.cpp

[Reid Spencer]

Changes in directory llvm/lib/Bytecode/Reader:

Reader.cpp updated: 1.214 -> 1.215
---
Log message:

For PR950: http://llvm.org/PR950 :
Update for signless integer types and parameter attribute implementation.
Of significant note:
  1. This changes the bytecode format yet again.
  2. There are 1/2 as many integer type planes (this is a good thing)
  3. GEP indices now use only 1 bit to identify their type which means 
 more GEP instructions won't be relegated to format 0 (size win)
  4. Parameter attributes are implemented but currently being stored
 verbosely for each function type. Some other day this needs to be
 optimized for size.


---
Diffs of the changes:  (+29 -34)

 Reader.cpp |   63 -
 1 files changed, 29 insertions(+), 34 deletions(-)


Index: llvm/lib/Bytecode/Reader/Reader.cpp
diff -u llvm/lib/Bytecode/Reader/Reader.cpp:1.214 
llvm/lib/Bytecode/Reader/Reader.cpp:1.215
--- llvm/lib/Bytecode/Reader/Reader.cpp:1.214   Fri Dec 15 13:49:23 2006
+++ llvm/lib/Bytecode/Reader/Reader.cpp Sat Dec 30 23:44:24 2006
@@ -44,7 +44,7 @@
 Use Op;
 ConstantPlaceHolder(const Type *Ty)
   : ConstantExpr(Ty, Instruction::UserOp1, &Op, 1),
-Op(UndefValue::get(Type::IntTy), this) {
+Op(UndefValue::get(Type::Int32Ty), this) {
 }
   };
 }
@@ -572,7 +572,7 @@
   if (Oprnds.size() != 2)
 error("Invalid extractelement instruction!");
   Value *V1 = getValue(iType, Oprnds[0]);
-  Value *V2 = getValue(Type::UIntTyID, Oprnds[1]);
+  Value *V2 = getValue(Type::Int32TyID, Oprnds[1]);
   
   if (!ExtractElementInst::isValidOperands(V1, V2))
 error("Invalid extractelement instruction!");
@@ -587,7 +587,7 @@
   
   Value *V1 = getValue(iType, Oprnds[0]);
   Value *V2 = getValue(getTypeSlot(PackedTy->getElementType()),Oprnds[1]);
-  Value *V3 = getValue(Type::UIntTyID, Oprnds[2]);
+  Value *V3 = getValue(Type::Int32TyID, Oprnds[2]);
 
   if (!InsertElementInst::isValidOperands(V1, V2, V3))
 error("Invalid insertelement instruction!");
@@ -601,7 +601,7 @@
   Value *V1 = getValue(iType, Oprnds[0]);
   Value *V2 = getValue(iType, Oprnds[1]);
   const PackedType *EltTy = 
-PackedType::get(Type::UIntTy, PackedTy->getNumElements());
+PackedType::get(Type::Int32Ty, PackedTy->getNumElements());
   Value *V3 = getValue(getTypeSlot(EltTy), Oprnds[2]);
   if (!ShuffleVectorInst::isValidOperands(V1, V2, V3))
 error("Invalid shufflevector instruction!");
@@ -713,7 +713,7 @@
 case Instruction::AShr:
   Result = new ShiftInst(Instruction::OtherOps(Opcode),
  getValue(iType, Oprnds[0]),
- getValue(Type::UByteTyID, Oprnds[1]));
+ getValue(Type::Int8TyID, Oprnds[1]));
   break;
 case Instruction::Ret:
   if (Oprnds.size() == 0)
@@ -876,7 +876,7 @@
 error("Invalid malloc instruction!");
 
   Result = new MallocInst(cast(InstTy)->getElementType(),
-  getValue(Type::UIntTyID, Oprnds[0]), Align);
+  getValue(Type::Int32TyID, Oprnds[0]), Align);
   break;
 }
 case Instruction::Alloca: {
@@ -889,7 +889,7 @@
 error("Invalid alloca instruction!");
 
   Result = new AllocaInst(cast(InstTy)->getElementType(),
-  getValue(Type::UIntTyID, Oprnds[0]), Align);
+  getValue(Type::Int32TyID, Oprnds[0]), Align);
   break;
 }
 case Instruction::Free:
@@ -913,18 +913,16 @@
 unsigned IdxTy = 0;
 // Struct indices are always uints, sequential type indices can be 
 // any of the 32 or 64-bit integer types.  The actual choice of 
-// type is encoded in the low two bits of the slot number.
+// type is encoded in the low bit of the slot number.
 if (isa(TopTy))
-  IdxTy = Type::UIntTyID;
+  IdxTy = Type::Int32TyID;
 else {
-  switch (ValIdx & 3) {
+  switch (ValIdx & 1) {
   default:
-  case 0: IdxTy = Type::UIntTyID; break;
-  case 1: IdxTy = Type::IntTyID; break;
-  case 2: IdxTy = Type::ULongTyID; break;
-  case 3: IdxTy = Type::LongTyID; break;
+  case 0: IdxTy = Type::Int32TyID; break;
+  case 1: IdxTy = Type::Int64TyID; break;
   }
-  ValIdx >>= 2;
+  ValIdx >>= 1;
 }
 Idx.push_back(getValue(IdxTy, ValIdx));
 NextTy = GetElementPtrInst::getIndexedType(InstTy, Idx, true);
@@ -1159,17 +1157,23 @@
   switch (PrimType) {
   case Type::FunctionTyID: {
 const Type *RetType = readType();
+unsigned RetAttr = read_vbr_uint();
 
 unsigned NumParams = read_vbr_uint();
 
 std::vector Params;
-while (NumParams--)
+std::vector Attrs;
+Attrs.push_back(FunctionType::ParameterAttrib

[llvm-commits] CVS: llvm/lib/Bytecode/Writer/SlotCalculator.cpp Writer.cpp

[Reid Spencer]

Changes in directory llvm/lib/Bytecode/Writer:

SlotCalculator.cpp updated: 1.76 -> 1.77
Writer.cpp updated: 1.140 -> 1.141
---
Log message:

For PR950: http://llvm.org/PR950 :
Update for signless integer types and parameter attribute implementation.
Of significant note:
  1. This changes the bytecode format yet again.
  2. There are 1/2 as many integer type planes (this is a good thing)
  3. GEP indices now use only 1 bit to identify their type which means 
 more GEP instructions won't be relegated to format 0 (size win)
  4. Parameter attributes are implemented but currently being stored
 verbosely for each function type. Some other day this needs to be
 optimized for size.


---
Diffs of the changes:  (+15 -23)

 SlotCalculator.cpp |3 +--
 Writer.cpp |   35 ++-
 2 files changed, 15 insertions(+), 23 deletions(-)


Index: llvm/lib/Bytecode/Writer/SlotCalculator.cpp
diff -u llvm/lib/Bytecode/Writer/SlotCalculator.cpp:1.76 
llvm/lib/Bytecode/Writer/SlotCalculator.cpp:1.77
--- llvm/lib/Bytecode/Writer/SlotCalculator.cpp:1.76Wed Dec  6 19:30:31 2006
+++ llvm/lib/Bytecode/Writer/SlotCalculator.cpp Sat Dec 30 23:44:24 2006
@@ -145,8 +145,7 @@
   //
   for (unsigned plane = 0, e = Table.size(); plane != e; ++plane) {
 if (const ArrayType *AT = dyn_cast(Types[plane]))
-  if (AT->getElementType() == Type::SByteTy ||
-  AT->getElementType() == Type::UByteTy) {
+  if (AT->getElementType() == Type::Int8Ty) {
 TypePlane &Plane = Table[plane];
 unsigned FirstNonStringID = 0;
 for (unsigned i = 0, e = Plane.size(); i != e; ++i)


Index: llvm/lib/Bytecode/Writer/Writer.cpp
diff -u llvm/lib/Bytecode/Writer/Writer.cpp:1.140 
llvm/lib/Bytecode/Writer/Writer.cpp:1.141
--- llvm/lib/Bytecode/Writer/Writer.cpp:1.140   Tue Dec 19 17:16:47 2006
+++ llvm/lib/Bytecode/Writer/Writer.cpp Sat Dec 30 23:44:24 2006
@@ -214,16 +214,20 @@
 int Slot = Table.getSlot(MT->getReturnType());
 assert(Slot != -1 && "Type used but not available!!");
 output_typeid((unsigned)Slot);
+output_vbr(unsigned(MT->getParamAttrs(0)));
 
 // Output the number of arguments to function (+1 if varargs):
 output_vbr((unsigned)MT->getNumParams()+MT->isVarArg());
 
 // Output all of the arguments...
 FunctionType::param_iterator I = MT->param_begin();
+unsigned Idx = 1;
 for (; I != MT->param_end(); ++I) {
   Slot = Table.getSlot(*I);
   assert(Slot != -1 && "Type used but not available!!");
   output_typeid((unsigned)Slot);
+  output_vbr(unsigned(MT->getParamAttrs(Idx)));
+  Idx++;
 }
 
 // Terminate list with VoidTy if we are a varargs function...
@@ -323,20 +327,13 @@
   output_vbr(0U);
 break;
 
-  case Type::UByteTyID:   // Unsigned integer types...
-  case Type::UShortTyID:
-  case Type::UIntTyID:
-  case Type::ULongTyID:
+  case Type::Int8TyID:   // Unsigned integer types...
+  case Type::Int16TyID:
+  case Type::Int32TyID:
+  case Type::Int64TyID:
 output_vbr(cast(CPV)->getZExtValue());
 break;
 
-  case Type::SByteTyID:   // Signed integer types...
-  case Type::ShortTyID:
-  case Type::IntTyID:
-  case Type::LongTyID:
-output_vbr(cast(CPV)->getSExtValue());
-break;
-
   case Type::ArrayTyID: {
 const ConstantArray *CPA = cast(CPV);
 assert(!CPA->isString() && "Constant strings should be handled 
specially!");
@@ -489,12 +486,10 @@
 unsigned IdxId;
 switch (I->getOperand(Idx)->getType()->getTypeID()) {
 default: assert(0 && "Unknown index type!");
-case Type::UIntTyID:  IdxId = 0; break;
-case Type::IntTyID:   IdxId = 1; break;
-case Type::ULongTyID: IdxId = 2; break;
-case Type::LongTyID:  IdxId = 3; break;
+case Type::Int32TyID:  IdxId = 0; break;
+case Type::Int64TyID:  IdxId = 1; break;
 }
-Slot = (Slot << 2) | IdxId;
+Slot = (Slot << 1) | IdxId;
   }
   output_vbr(unsigned(Slot));
 }
@@ -742,12 +737,10 @@
   unsigned IdxId;
   switch (GEP->getOperand(Idx)->getType()->getTypeID()) {
   default: assert(0 && "Unknown index type!");
-  case Type::UIntTyID:  IdxId = 0; break;
-  case Type::IntTyID:   IdxId = 1; break;
-  case Type::ULongTyID: IdxId = 2; break;
-  case Type::LongTyID:  IdxId = 3; break;
+  case Type::Int32TyID: IdxId = 0; break;
+  case Type::Int64TyID: IdxId = 1; break;
   }
-  Slots[Idx] = (Slots[Idx] << 2) | IdxId;
+  Slots[Idx] = (Slots[Idx] << 1) | IdxId;
   if (Slots[Idx] > MaxOpSlot) MaxOpSlot = Slots[Idx];
 }
 } else if (Opcode == 58) {



___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

[llvm-commits] CVS: llvm/tools/llvm-upgrade/UpgradeLexer.l UpgradeParser.y

[Reid Spencer]

Changes in directory llvm/tools/llvm-upgrade:

UpgradeLexer.l updated: 1.10 -> 1.11
UpgradeParser.y updated: 1.26 -> 1.27
---
Log message:

For PR950: http://llvm.org/PR950 :
Don't attempt to parse both the old and new grammars. It is near impossible
to get it right. Remove support for the new define keyword and don't 
attempt to insert parameter attributes because there isn't enough
contextual information for it.


---
Diffs of the changes:  (+29 -30)

 UpgradeLexer.l  |   21 -
 UpgradeParser.y |   38 +-
 2 files changed, 29 insertions(+), 30 deletions(-)


Index: llvm/tools/llvm-upgrade/UpgradeLexer.l
diff -u llvm/tools/llvm-upgrade/UpgradeLexer.l:1.10 
llvm/tools/llvm-upgrade/UpgradeLexer.l:1.11
--- llvm/tools/llvm-upgrade/UpgradeLexer.l:1.10 Fri Dec 29 14:33:37 2006
+++ llvm/tools/llvm-upgrade/UpgradeLexer.l  Sat Dec 30 23:45:57 2006
@@ -105,7 +105,6 @@
 true{ RET_TOK( TRUETOK);  }
 false   { RET_TOK( FALSETOK); }
 declare { RET_TOK( DECLARE); }
-define  { RET_TOK( DEFINE); }
 global  { RET_TOK( GLOBAL); }
 constant{ RET_TOK( CONSTANT); }
 internal{ RET_TOK( INTERNAL); }
@@ -149,14 +148,18 @@
 
 void{ RET_TY(VOID,VoidTy,"void",false); }
 bool{ RET_TY(BOOL,BoolTy,"bool",false); }
-sbyte   { RET_TY(SBYTE,SByteTy,"sbyte",true); }
-ubyte   { RET_TY(UBYTE,UByteTy,"ubyte",false); }
-short   { RET_TY(SHORT,ShortTy,"short",true); }
-ushort  { RET_TY(USHORT,UShortTy,"ushort",false); }
-int { RET_TY(INT,IntTy,"int",true);   }
-uint{ RET_TY(UINT,UIntTy,"uint",false);  }
-long{ RET_TY(LONG,LongTy,"long",true);  }
-ulong   { RET_TY(ULONG,ULongTy,"ulong",false); }
+sbyte   { RET_TY(SBYTE,SByteTy,"i8",true); }
+ubyte   { RET_TY(UBYTE,UByteTy,"i8",false); }
+short   { RET_TY(SHORT,ShortTy,"i16",true); }
+ushort  { RET_TY(USHORT,UShortTy,"i16",false); }
+int { RET_TY(INT,IntTy,"i32",true);   }
+uint{ RET_TY(UINT,UIntTy,"i32",false);  }
+long{ RET_TY(LONG,LongTy,"i64",true);  }
+ulong   { RET_TY(ULONG,ULongTy,"i64",false); }
+i8  { RET_TY(UBYTE,UByteTy,"i8",false); }
+i16 { RET_TY(USHORT,UShortTy,"i16",false); }
+i32 { RET_TY(UINT,UIntTy,"i32",false); }
+i64 { RET_TY(ULONG,ULongTy,"i64",false); }
 float   { RET_TY(FLOAT,FloatTy,"float",false); }
 double  { RET_TY(DOUBLE,DoubleTy,"double",false); }
 label   { RET_TY(LABEL,LabelTy,"label",false); }


Index: llvm/tools/llvm-upgrade/UpgradeParser.y
diff -u llvm/tools/llvm-upgrade/UpgradeParser.y:1.26 
llvm/tools/llvm-upgrade/UpgradeParser.y:1.27
--- llvm/tools/llvm-upgrade/UpgradeParser.y:1.26Fri Dec 29 14:33:37 2006
+++ llvm/tools/llvm-upgrade/UpgradeParser.y Sat Dec 30 23:45:57 2006
@@ -34,6 +34,10 @@
 unsigned SizeOfPointer = 32;
 static uint64_t unique = 1;
 
+// This bool controls whether attributes are ever added to function 
declarations
+// definitions and calls.
+static bool AddAttributes = false;
+
 typedef std::vector TypeVector;
 static TypeVector EnumeratedTypes;
 typedef std::map TypeMap;
@@ -50,12 +54,13 @@
 }
 
 void UpgradeAssembly(const std::string &infile, std::istream& in, 
- std::ostream &out, bool debug)
+ std::ostream &out, bool debug, bool addAttrs)
 {
   Upgradelineno = 1; 
   CurFilename = infile;
   LexInput = ∈
   yydebug = debug;
+  AddAttributes = addAttrs;
   O = &out;
 
   if (yyparse()) {
@@ -176,15 +181,15 @@
 // fp -> ptr cast is no longer supported but we must upgrade this
 // by doing a double cast: fp -> int -> ptr
 if (isConst)
-  Source = "ulong fptoui(" + Source + " to ulong)";
+  Source = "i64 fptoui(" + Source + " to i64)";
 else {
   *O << "%cast_upgrade" << unique << " = fptoui " << Source 
- << " to ulong\n";
-  Source = "ulong %cast_upgrade" + llvm::utostr(unique);
+ << " to i64\n";
+  Source = "i64 %cast_upgrade" + llvm::utostr(unique);
 }
 // Update the SrcTy for the getCastOpcode call below
 SrcTy.destroy();
-SrcTy.newTy = new std::string("ulong");
+SrcTy.newTy = new std::string("i64");
 SrcTy.oldTy = ULongTy;
   } else if (DstTy.oldTy == BoolTy && SrcTy.oldTy != BoolTy) {
 // cast ptr %x to  bool was previously defined as setne ptr %x, null
@@ -286,7 +291,7 @@
 %token  NULL_TOK UNDEF ZEROINITIALIZER TRUETOK FALSETOK
 %token  TYPE VAR_ID LABELSTR STRINGCONSTANT
 %token  IMPLEMENTATION BEGINTOK ENDTOK
-%token  DECLARE DEFINE GLOBAL CONSTANT SECTION VOLATILE
+%token  DECLARE GLOBAL CONSTANT SECTION VOLATILE
 %token  TO DOTDOTDOT CONST INTERNAL LINKONCE WEAK 
 %token  DLLIMPORT DLLEXPORT EXTERN_WEAK APPENDING
 %token  NOT EXTERNAL TARGET TRIPLE ENDIAN POINTERSIZE LITTLE BIG
@@ -922,8 +927,8 @@
   }
   | /* empty */ { $$ = new std::

[llvm-commits] CVS: llvm/lib/Analysis/BasicAliasAnalysis.cpp ConstantFolding.cpp ScalarEvolution.cpp

[Reid Spencer]

Changes in directory llvm/lib/Analysis:

BasicAliasAnalysis.cpp updated: 1.95 -> 1.96
ConstantFolding.cpp updated: 1.7 -> 1.8
ScalarEvolution.cpp updated: 1.77 -> 1.78
---
Log message:

For PR950: http://llvm.org/PR950 :
This patch replaces signed integer types with signless ones:
1. [US]Byte -> Int8
2. [U]Short -> Int16
3. [U]Int   -> Int32
4. [U]Long  -> Int64.
5. Removal of isSigned, isUnsigned, getSignedVersion, getUnsignedVersion
   and other methods related to signedness. In a few places this warranted
   identifying the signedness information from other sources.



---
Diffs of the changes:  (+21 -68)

 BasicAliasAnalysis.cpp |   32 ++--
 ConstantFolding.cpp|1 
 ScalarEvolution.cpp|   56 +++--
 3 files changed, 21 insertions(+), 68 deletions(-)


Index: llvm/lib/Analysis/BasicAliasAnalysis.cpp
diff -u llvm/lib/Analysis/BasicAliasAnalysis.cpp:1.95 
llvm/lib/Analysis/BasicAliasAnalysis.cpp:1.96
--- llvm/lib/Analysis/BasicAliasAnalysis.cpp:1.95   Sat Dec 23 00:05:40 2006
+++ llvm/lib/Analysis/BasicAliasAnalysis.cppSat Dec 30 23:48:39 2006
@@ -269,7 +269,7 @@
   if (V1 == V2) return MustAlias;
 
   if ((!isa(V1->getType()) || !isa(V2->getType())) &&
-  V1->getType() != Type::LongTy && V2->getType() != Type::LongTy)
+  V1->getType() != Type::Int64Ty && V2->getType() != Type::Int64Ty)
 return NoAlias;  // Scalars cannot alias each other
 
   // Strip off cast instructions...
@@ -458,14 +458,10 @@
   if (Constant *C1 = dyn_cast(V1))
 if (Constant *C2 = dyn_cast(V2)) {
   // Sign extend the constants to long types, if necessary
-  if (C1->getType()->getPrimitiveSizeInBits() < 64)
-C1 = ConstantExpr::getSExt(C1, Type::LongTy);
-  else if (C1->getType() == Type::ULongTy)
-C1 = ConstantExpr::getBitCast(C1, Type::LongTy);
-  if (C2->getType()->getPrimitiveSizeInBits() < 64)
-C2 = ConstantExpr::getSExt(C2, Type::LongTy);
-  else if (C2->getType() == Type::ULongTy)
-C2 = ConstantExpr::getBitCast(C2, Type::LongTy);
+  if (C1->getType() != Type::Int64Ty)
+C1 = ConstantExpr::getSExt(C1, Type::Int64Ty);
+  if (C2->getType() != Type::Int64Ty) 
+C2 = ConstantExpr::getSExt(C2, Type::Int64Ty);
   return C1 == C2;
 }
   return false;
@@ -554,14 +550,10 @@
 if (Constant *G2OC = 
dyn_cast(const_cast(G2Oper))){
   if (G1OC->getType() != G2OC->getType()) {
 // Sign extend both operands to long.
-if (G1OC->getType()->getPrimitiveSizeInBits() < 64)
-  G1OC = ConstantExpr::getSExt(G1OC, Type::LongTy);
-else if (G1OC->getType() == Type::ULongTy)
-  G1OC = ConstantExpr::getBitCast(G1OC, Type::LongTy);
-if (G2OC->getType()->getPrimitiveSizeInBits() < 64)
-  G2OC = ConstantExpr::getSExt(G2OC, Type::LongTy);
-else if (G2OC->getType() == Type::ULongTy)
-  G2OC = ConstantExpr::getBitCast(G2OC, Type::LongTy);
+if (G1OC->getType() != Type::Int64Ty)
+  G1OC = ConstantExpr::getSExt(G1OC, Type::Int64Ty);
+if (G2OC->getType() != Type::Int64Ty) 
+  G2OC = ConstantExpr::getSExt(G2OC, Type::Int64Ty);
 GEP1Ops[FirstConstantOper] = G1OC;
 GEP2Ops[FirstConstantOper] = G2OC;
   }
@@ -661,7 +653,7 @@
   const Type *ZeroIdxTy = GEPPointerTy;
   for (unsigned i = 0; i != FirstConstantOper; ++i) {
 if (!isa(ZeroIdxTy))
-  GEP1Ops[i] = GEP2Ops[i] = Constant::getNullValue(Type::UIntTy);
+  GEP1Ops[i] = GEP2Ops[i] = Constant::getNullValue(Type::Int32Ty);
 
 if (const CompositeType *CT = dyn_cast(ZeroIdxTy))
   ZeroIdxTy = CT->getTypeAtIndex(GEP1Ops[i]);
@@ -702,9 +694,9 @@
   // value possible.
   //
   if (const ArrayType *AT = dyn_cast(BasePtr1Ty))
-GEP1Ops[i] = ConstantInt::get(Type::LongTy, 
AT->getNumElements()-1);
+GEP1Ops[i] = ConstantInt::get(Type::Int64Ty, 
AT->getNumElements()-1);
   else if (const PackedType *PT = dyn_cast(BasePtr1Ty))
-GEP1Ops[i] = ConstantInt::get(Type::LongTy, 
PT->getNumElements()-1);
+GEP1Ops[i] = ConstantInt::get(Type::Int64Ty, 
PT->getNumElements()-1);
 
 }
   }


Index: llvm/lib/Analysis/ConstantFolding.cpp
diff -u llvm/lib/Analysis/ConstantFolding.cpp:1.7 
llvm/lib/Analysis/ConstantFolding.cpp:1.8
--- llvm/lib/Analysis/ConstantFolding.cpp:1.7   Fri Dec  1 20:22:01 2006
+++ llvm/lib/Analysis/ConstantFolding.cpp   Sat Dec 30 23:48:39 2006
@@ -164,7 +164,6 @@
   break;
   }
 } else if (ConstantInt *Op = dyn_cast(Operands[0])) {
-  assert(Op->getType()->isUnsigned() && "bswap args must be unsigned");
   uint64_t V = Op->getZExtValue();
   if (Name == "llvm.bswap.i16")
 return ConstantInt::get(Ty, ByteSwap_16(V));


Index: llvm/lib/Analysis/ScalarEvolution.cpp
diff -u llvm/lib/Analysis/Sc

[llvm-commits] CVS: llvm/lib/Transforms/ExprTypeConvert.cpp LevelRaise.cpp TransformInternals.cpp

[Reid Spencer]

Changes in directory llvm/lib/Transforms:

ExprTypeConvert.cpp updated: 1.118 -> 1.119
LevelRaise.cpp updated: 1.117 -> 1.118
TransformInternals.cpp updated: 1.51 -> 1.52
---
Log message:

For PR950: http://llvm.org/PR950 :
This patch replaces signed integer types with signless ones:
1. [US]Byte -> Int8
2. [U]Short -> Int16
3. [U]Int   -> Int32
4. [U]Long  -> Int64.
5. Removal of isSigned, isUnsigned, getSignedVersion, getUnsignedVersion
   and other methods related to signedness. In a few places this warranted
   identifying the signedness information from other sources.



---
Diffs of the changes:  (+15 -16)

 ExprTypeConvert.cpp|   19 +--
 LevelRaise.cpp |4 ++--
 TransformInternals.cpp |8 
 3 files changed, 15 insertions(+), 16 deletions(-)


Index: llvm/lib/Transforms/ExprTypeConvert.cpp
diff -u llvm/lib/Transforms/ExprTypeConvert.cpp:1.118 
llvm/lib/Transforms/ExprTypeConvert.cpp:1.119
--- llvm/lib/Transforms/ExprTypeConvert.cpp:1.118   Sat Dec 23 00:05:41 2006
+++ llvm/lib/Transforms/ExprTypeConvert.cpp Sat Dec 30 23:48:39 2006
@@ -210,8 +210,9 @@
 Constant *CPV = cast(V);
 // Constants are converted by constant folding the cast that is required.
 // We assume here that all casts are implemented for constant prop.
-Instruction::CastOps opcode = CastInst::getCastOpcode(CPV,
-CPV->getType()->isSigned(), Ty, Ty->isSigned());
+// FIXME: This seems to work, but it is unclear why ZEXT is always the
+// right choice here.
+Instruction::CastOps opcode = CastInst::getCastOpcode(CPV, false, 
Ty,false);
 Value *Result = ConstantExpr::getCast(opcode, CPV, Ty);
 // Add the instruction to the expression map
 //VMC.ExprMap[V] = Result;
@@ -231,7 +232,7 @@
   case Instruction::BitCast: {
 assert(VMC.NewCasts.count(ValueHandle(VMC, I)) == 0);
 Instruction::CastOps opcode = CastInst::getCastOpcode(I->getOperand(0),
-I->getOperand(0)->getType()->isSigned(), Ty, Ty->isSigned());
+false, Ty, false);
 Res = CastInst::create(opcode, I->getOperand(0), Ty, Name);
 VMC.NewCasts.insert(ValueHandle(VMC, Res));
 break;
@@ -473,8 +474,6 @@
   }
   case Instruction::LShr:
   case Instruction::AShr:
-if (Ty->isSigned() != V->getType()->isSigned()) return false;
-// FALL THROUGH
   case Instruction::Shl:
 if (I->getOperand(1) == V) return false;  // Cannot change shift amount 
type
 if (!Ty->isInteger()) return false;
@@ -713,8 +712,8 @@
 
   switch (I->getOpcode()) {
   case Instruction::BitCast: {
-Instruction::CastOps opcode = CastInst::getCastOpcode(NewVal,
-NewVal->getType()->isSigned(), I->getType(), I->getType()->isSigned());
+Instruction::CastOps opcode = CastInst::getCastOpcode(NewVal, false, 
+  I->getType(), false);
 Res = CastInst::create(opcode, NewVal, I->getType(), Name);
 break;
   }
@@ -768,7 +767,7 @@
 
 if (const CompositeType *CT = dyn_cast(LoadedTy)) {
   std::vector Indices;
-  Indices.push_back(Constant::getNullValue(Type::UIntTy));
+  Indices.push_back(Constant::getNullValue(Type::Int32Ty));
 
   unsigned Offset = 0;   // No offset, get first leaf.
   LoadedTy = getStructOffsetType(CT, Offset, Indices, TD, false);
@@ -801,7 +800,7 @@
 
 if (ElTy != NewTy) {
   std::vector Indices;
-  Indices.push_back(Constant::getNullValue(Type::UIntTy));
+  Indices.push_back(Constant::getNullValue(Type::Int32Ty));
 
   unsigned Offset = 0;
   const Type *Ty = getStructOffsetType(ElTy, Offset, Indices, 
TD,false);
@@ -830,7 +829,7 @@
 
   if (isa(ValTy)) {
 std::vector Indices;
-Indices.push_back(Constant::getNullValue(Type::UIntTy));
+Indices.push_back(Constant::getNullValue(Type::Int32Ty));
 
 unsigned Offset = 0;
 ValTy = getStructOffsetType(ValTy, Offset, Indices, TD, false);


Index: llvm/lib/Transforms/LevelRaise.cpp
diff -u llvm/lib/Transforms/LevelRaise.cpp:1.117 
llvm/lib/Transforms/LevelRaise.cpp:1.118
--- llvm/lib/Transforms/LevelRaise.cpp:1.117Sat Dec 23 00:05:41 2006
+++ llvm/lib/Transforms/LevelRaise.cpp  Sat Dec 30 23:48:39 2006
@@ -206,7 +206,7 @@
   // Build the index vector, full of all zeros
   std::vector Indices;
 
-  Indices.push_back(Constant::getNullValue(Type::UIntTy));
+  Indices.push_back(Constant::getNullValue(Type::Int32Ty));
   while (CurCTy && !isa(CurCTy)) {
 if (const StructType *CurSTy = dyn_cast(CurCTy)) {
   // Check for a zero element struct type... if we have one, bail.
@@ -221,7 +221,7 @@
 }
 
 // Insert a zero to index through this type...
-Indices.push_back(Constant::getNullValue(Type::UIntTy));
+Indices.push_back(Constant::getNullValue(Type::Int32Ty));
 
 // Did we find what we're looking for?
 if (ElT

[llvm-commits] CVS: llvm/lib/Transforms/Scalar/CorrelatedExprs.cpp IndVarSimplify.cpp InstructionCombining.cpp LoopStrengthReduce.cpp LowerGC.cpp LowerPacked.cpp ScalarReplAggregates.cpp

[Reid Spencer]

Changes in directory llvm/lib/Transforms/Scalar:

CorrelatedExprs.cpp updated: 1.42 -> 1.43
IndVarSimplify.cpp updated: 1.99 -> 1.100
InstructionCombining.cpp updated: 1.576 -> 1.577
LoopStrengthReduce.cpp updated: 1.102 -> 1.103
LowerGC.cpp updated: 1.16 -> 1.17
LowerPacked.cpp updated: 1.14 -> 1.15
ScalarReplAggregates.cpp updated: 1.61 -> 1.62
---
Log message:

For PR950: http://llvm.org/PR950 :
This patch replaces signed integer types with signless ones:
1. [US]Byte -> Int8
2. [U]Short -> Int16
3. [U]Int   -> Int32
4. [U]Long  -> Int64.
5. Removal of isSigned, isUnsigned, getSignedVersion, getUnsignedVersion
   and other methods related to signedness. In a few places this warranted
   identifying the signedness information from other sources.



---
Diffs of the changes:  (+158 -227)

 CorrelatedExprs.cpp  |2 
 IndVarSimplify.cpp   |5 
 InstructionCombining.cpp |  291 ++-
 LoopStrengthReduce.cpp   |   12 -
 LowerGC.cpp  |   18 +-
 LowerPacked.cpp  |   14 +-
 ScalarReplAggregates.cpp |   43 +++---
 7 files changed, 158 insertions(+), 227 deletions(-)


Index: llvm/lib/Transforms/Scalar/CorrelatedExprs.cpp
diff -u llvm/lib/Transforms/Scalar/CorrelatedExprs.cpp:1.42 
llvm/lib/Transforms/Scalar/CorrelatedExprs.cpp:1.43
--- llvm/lib/Transforms/Scalar/CorrelatedExprs.cpp:1.42 Sat Dec 23 00:05:41 2006
+++ llvm/lib/Transforms/Scalar/CorrelatedExprs.cpp  Sat Dec 30 23:48:39 2006
@@ -111,7 +111,7 @@
 Value *Replacement;
   public:
 ValueInfo(const Type *Ty)
-  : Bounds(Ty->isIntegral() ? Ty : Type::IntTy), Replacement(0) {}
+  : Bounds(Ty->isIntegral() ? Ty : Type::Int32Ty), Replacement(0) {}
 
 // getBounds() - Return the constant bounds of the value...
 const ConstantRange &getBounds() const { return Bounds; }


Index: llvm/lib/Transforms/Scalar/IndVarSimplify.cpp
diff -u llvm/lib/Transforms/Scalar/IndVarSimplify.cpp:1.99 
llvm/lib/Transforms/Scalar/IndVarSimplify.cpp:1.100
--- llvm/lib/Transforms/Scalar/IndVarSimplify.cpp:1.99  Sat Dec 23 00:05:41 2006
+++ llvm/lib/Transforms/Scalar/IndVarSimplify.cpp   Sat Dec 30 23:48:39 2006
@@ -173,7 +173,7 @@
   Constant *NCE = ConstantExpr::getGetElementPtr(CE->getOperand(0),
  CEIdxs);
   GetElementPtrInst *NGEPI =
-new GetElementPtrInst(NCE, Constant::getNullValue(Type::IntTy),
+new GetElementPtrInst(NCE, 
Constant::getNullValue(Type::Int32Ty),
   NewAdd, GEPI->getName(), GEPI);
   GEPI->replaceAllUsesWith(NGEPI);
   GEPI->eraseFromParent();
@@ -499,7 +499,6 @@
 
   // Now that we know the largest of of the induction variables in this loop,
   // insert a canonical induction variable of the largest size.
-  LargestType = LargestType->getUnsignedVersion();
   Value *IndVar = 
Rewriter.getOrInsertCanonicalInductionVariable(L,LargestType);
   ++NumInserted;
   Changed = true;
@@ -525,7 +524,7 @@
 PHINode *PN = IndVars[i].first;
 InsertedSizes[PN->getType()->getPrimitiveSize()] = true;
 Instruction *New = CastInst::create(Instruction::Trunc, IndVar, 
-PN->getType()->getUnsignedVersion(), "indvar", InsertPt);
+PN->getType(), "indvar", InsertPt);
 Rewriter.addInsertedValue(New, SE->getSCEV(New));
   }
   }


Index: llvm/lib/Transforms/Scalar/InstructionCombining.cpp
diff -u llvm/lib/Transforms/Scalar/InstructionCombining.cpp:1.576 
llvm/lib/Transforms/Scalar/InstructionCombining.cpp:1.577
--- llvm/lib/Transforms/Scalar/InstructionCombining.cpp:1.576   Sat Dec 23 
18:40:59 2006
+++ llvm/lib/Transforms/Scalar/InstructionCombining.cpp Sat Dec 30 23:48:39 2006
@@ -340,10 +340,8 @@
 // though a va_arg area...
 static const Type *getPromotedType(const Type *Ty) {
   switch (Ty->getTypeID()) {
-  case Type::SByteTyID:
-  case Type::ShortTyID:  return Type::IntTy;
-  case Type::UByteTyID:
-  case Type::UShortTyID: return Type::UIntTy;
+  case Type::Int8TyID:
+  case Type::Int16TyID:  return Type::Int32Ty;
   case Type::FloatTyID:  return Type::DoubleTy;
   default:   return Ty;
   }
@@ -540,14 +538,9 @@
 /// GetConstantInType - Return a ConstantInt with the specified type and value.
 ///
 static ConstantIntegral *GetConstantInType(const Type *Ty, uint64_t Val) {
-  if (Ty->isUnsigned()) 
-return ConstantInt::get(Ty, Val);
-  else if (Ty->getTypeID() == Type::BoolTyID)
+  if (Ty->getTypeID() == Type::BoolTyID)
 return ConstantBool::get(Val);
-  int64_t SVal = Val;
-  SVal <<= 64-Ty->getPrimitiveSizeInBits();
-  SVal >>= 64-Ty->getPrimitiveSizeInBits();
-  return ConstantInt::get(Ty, SVal);
+  return ConstantInt::get(Ty, Val);
 }
 
 
@@ -1575,7 +1568,7 @@
   bool shouldApply(Value *LHS) const { return LHS == RHS; }
   Instruction *apply(BinaryOperator &Add) const {
 return new ShiftInst(Instruction::Shl, Add.getOpera

[llvm-commits] CVS: llvm/lib/Transforms/IPO/ArgumentPromotion.cpp DeadArgumentElimination.cpp GlobalOpt.cpp IndMemRemoval.cpp LowerSetJmp.cpp RaiseAllocations.cpp SimplifyLibCalls.cpp

[Reid Spencer]

Changes in directory llvm/lib/Transforms/IPO:

ArgumentPromotion.cpp updated: 1.32 -> 1.33
DeadArgumentElimination.cpp updated: 1.33 -> 1.34
GlobalOpt.cpp updated: 1.81 -> 1.82
IndMemRemoval.cpp updated: 1.9 -> 1.10
LowerSetJmp.cpp updated: 1.34 -> 1.35
RaiseAllocations.cpp updated: 1.34 -> 1.35
SimplifyLibCalls.cpp updated: 1.80 -> 1.81
---
Log message:

For PR950: http://llvm.org/PR950 :
This patch replaces signed integer types with signless ones:
1. [US]Byte -> Int8
2. [U]Short -> Int16
3. [U]Int   -> Int32
4. [U]Long  -> Int64.
5. Removal of isSigned, isUnsigned, getSignedVersion, getUnsignedVersion
   and other methods related to signedness. In a few places this warranted
   identifying the signedness information from other sources.



---
Diffs of the changes:  (+133 -135)

 ArgumentPromotion.cpp   |6 -
 DeadArgumentElimination.cpp |4 
 GlobalOpt.cpp   |   17 +--
 IndMemRemoval.cpp   |4 
 LowerSetJmp.cpp |   24 ++---
 RaiseAllocations.cpp|   22 ++---
 SimplifyLibCalls.cpp|  191 +---
 7 files changed, 133 insertions(+), 135 deletions(-)


Index: llvm/lib/Transforms/IPO/ArgumentPromotion.cpp
diff -u llvm/lib/Transforms/IPO/ArgumentPromotion.cpp:1.32 
llvm/lib/Transforms/IPO/ArgumentPromotion.cpp:1.33
--- llvm/lib/Transforms/IPO/ArgumentPromotion.cpp:1.32  Tue Dec 19 16:09:18 2006
+++ llvm/lib/Transforms/IPO/ArgumentPromotion.cpp   Sat Dec 30 23:48:39 2006
@@ -384,7 +384,7 @@
   bool ExtraArgHack = false;
   if (Params.empty() && FTy->isVarArg()) {
 ExtraArgHack = true;
-Params.push_back(Type::IntTy);
+Params.push_back(Type::Int32Ty);
   }
   FunctionType *NFTy = FunctionType::get(RetTy, Params, FTy->isVarArg());
 
@@ -429,7 +429,7 @@
   }
 
 if (ExtraArgHack)
-  Args.push_back(Constant::getNullValue(Type::IntTy));
+  Args.push_back(Constant::getNullValue(Type::Int32Ty));
 
 // Push any varargs arguments on the list
 for (; AI != CS.arg_end(); ++AI)
@@ -540,7 +540,7 @@
 
   // Notify the alias analysis implementation that we inserted a new argument.
   if (ExtraArgHack)
-AA.copyValue(Constant::getNullValue(Type::IntTy), NF->arg_begin());
+AA.copyValue(Constant::getNullValue(Type::Int32Ty), NF->arg_begin());
 
 
   // Tell the alias analysis that the old function is about to disappear.


Index: llvm/lib/Transforms/IPO/DeadArgumentElimination.cpp
diff -u llvm/lib/Transforms/IPO/DeadArgumentElimination.cpp:1.33 
llvm/lib/Transforms/IPO/DeadArgumentElimination.cpp:1.34
--- llvm/lib/Transforms/IPO/DeadArgumentElimination.cpp:1.33Tue Dec 19 
16:09:18 2006
+++ llvm/lib/Transforms/IPO/DeadArgumentElimination.cpp Sat Dec 30 23:48:39 2006
@@ -500,7 +500,7 @@
   bool ExtraArgHack = false;
   if (Params.empty() && FTy->isVarArg()) {
 ExtraArgHack = true;
-Params.push_back(Type::IntTy);
+Params.push_back(Type::Int32Ty);
   }
 
   FunctionType *NFTy = FunctionType::get(RetTy, Params, FTy->isVarArg());
@@ -526,7 +526,7 @@
 Args.push_back(*AI);
 
 if (ExtraArgHack)
-  Args.push_back(UndefValue::get(Type::IntTy));
+  Args.push_back(UndefValue::get(Type::Int32Ty));
 
 // Push any varargs arguments on the list
 for (; AI != CS.arg_end(); ++AI)


Index: llvm/lib/Transforms/IPO/GlobalOpt.cpp
diff -u llvm/lib/Transforms/IPO/GlobalOpt.cpp:1.81 
llvm/lib/Transforms/IPO/GlobalOpt.cpp:1.82
--- llvm/lib/Transforms/IPO/GlobalOpt.cpp:1.81  Sat Dec 23 00:05:41 2006
+++ llvm/lib/Transforms/IPO/GlobalOpt.cpp   Sat Dec 30 23:48:39 2006
@@ -384,7 +384,7 @@
 NewGlobals.reserve(STy->getNumElements());
 for (unsigned i = 0, e = STy->getNumElements(); i != e; ++i) {
   Constant *In = getAggregateConstantElement(Init,
-ConstantInt::get(Type::UIntTy, i));
+ConstantInt::get(Type::Int32Ty, 
i));
   assert(In && "Couldn't get element of initializer?");
   GlobalVariable *NGV = new GlobalVariable(STy->getElementType(i), false,
GlobalVariable::InternalLinkage,
@@ -406,7 +406,7 @@
 NewGlobals.reserve(NumElements);
 for (unsigned i = 0, e = NumElements; i != e; ++i) {
   Constant *In = getAggregateConstantElement(Init,
-ConstantInt::get(Type::UIntTy, i));
+ConstantInt::get(Type::Int32Ty, 
i));
   assert(In && "Couldn't get element of initializer?");
 
   GlobalVariable *NGV = new GlobalVariable(STy->getElementType(), false,
@@ -422,7 +422,7 @@
 
   DOUT << "PERFORMING GLOBAL SRA ON: " << *GV;
 
-  Constant *NullInt = Constant::getNullValue(Type::IntTy);
+  Constant *NullInt = Constant::getNullValue(Type::Int32Ty);
 
   // Loop over all of the uses of the global, replacing the constantexpr geps,
   // with smaller constantexpr geps or direct references.
@@ -679,10 +679,10 @@
 Type *NewTy =

[llvm-commits] CVS: llvm/lib/Transforms/Instrumentation/BlockProfiling.cpp EdgeProfiling.cpp EmitFunctions.cpp ProfilingUtils.cpp RSProfiling.cpp TraceBasicBlocks.cpp TraceValues.cpp

[Reid Spencer]

Changes in directory llvm/lib/Transforms/Instrumentation:

BlockProfiling.cpp updated: 1.18 -> 1.19
EdgeProfiling.cpp updated: 1.9 -> 1.10
EmitFunctions.cpp updated: 1.26 -> 1.27
ProfilingUtils.cpp updated: 1.12 -> 1.13
RSProfiling.cpp updated: 1.13 -> 1.14
TraceBasicBlocks.cpp updated: 1.19 -> 1.20
TraceValues.cpp updated: 1.78 -> 1.79
---
Log message:

For PR950: http://llvm.org/PR950 :
This patch replaces signed integer types with signless ones:
1. [US]Byte -> Int8
2. [U]Short -> Int16
3. [U]Int   -> Int32
4. [U]Long  -> Int64.
5. Removal of isSigned, isUnsigned, getSignedVersion, getUnsignedVersion
   and other methods related to signedness. In a few places this warranted
   identifying the signedness information from other sources.



---
Diffs of the changes:  (+37 -37)

 BlockProfiling.cpp   |4 ++--
 EdgeProfiling.cpp|2 +-
 EmitFunctions.cpp|8 
 ProfilingUtils.cpp   |   26 +-
 RSProfiling.cpp  |   16 
 TraceBasicBlocks.cpp |4 ++--
 TraceValues.cpp  |   14 +++---
 7 files changed, 37 insertions(+), 37 deletions(-)


Index: llvm/lib/Transforms/Instrumentation/BlockProfiling.cpp
diff -u llvm/lib/Transforms/Instrumentation/BlockProfiling.cpp:1.18 
llvm/lib/Transforms/Instrumentation/BlockProfiling.cpp:1.19
--- llvm/lib/Transforms/Instrumentation/BlockProfiling.cpp:1.18 Wed Dec  6 
19:30:31 2006
+++ llvm/lib/Transforms/Instrumentation/BlockProfiling.cpp  Sat Dec 30 
23:48:39 2006
@@ -57,7 +57,7 @@
 if (!I->isExternal())
   ++NumFunctions;
 
-  const Type *ATy = ArrayType::get(Type::UIntTy, NumFunctions);
+  const Type *ATy = ArrayType::get(Type::Int32Ty, NumFunctions);
   GlobalVariable *Counters =
 new GlobalVariable(ATy, false, GlobalValue::InternalLinkage,
Constant::getNullValue(ATy), "FuncProfCounters", &M);
@@ -99,7 +99,7 @@
   for (Module::iterator I = M.begin(), E = M.end(); I != E; ++I)
 NumBlocks += I->size();
 
-  const Type *ATy = ArrayType::get(Type::UIntTy, NumBlocks);
+  const Type *ATy = ArrayType::get(Type::Int32Ty, NumBlocks);
   GlobalVariable *Counters =
 new GlobalVariable(ATy, false, GlobalValue::InternalLinkage,
Constant::getNullValue(ATy), "BlockProfCounters", &M);


Index: llvm/lib/Transforms/Instrumentation/EdgeProfiling.cpp
diff -u llvm/lib/Transforms/Instrumentation/EdgeProfiling.cpp:1.9 
llvm/lib/Transforms/Instrumentation/EdgeProfiling.cpp:1.10
--- llvm/lib/Transforms/Instrumentation/EdgeProfiling.cpp:1.9   Wed Dec  6 
19:30:31 2006
+++ llvm/lib/Transforms/Instrumentation/EdgeProfiling.cpp   Sat Dec 30 
23:48:39 2006
@@ -58,7 +58,7 @@
   NumEdges += BB->getTerminator()->getNumSuccessors();
 }
 
-  const Type *ATy = ArrayType::get(Type::UIntTy, NumEdges);
+  const Type *ATy = ArrayType::get(Type::Int32Ty, NumEdges);
   GlobalVariable *Counters =
 new GlobalVariable(ATy, false, GlobalValue::InternalLinkage,
Constant::getNullValue(ATy), "EdgeProfCounters", &M);


Index: llvm/lib/Transforms/Instrumentation/EmitFunctions.cpp
diff -u llvm/lib/Transforms/Instrumentation/EmitFunctions.cpp:1.26 
llvm/lib/Transforms/Instrumentation/EmitFunctions.cpp:1.27
--- llvm/lib/Transforms/Instrumentation/EmitFunctions.cpp:1.26  Fri Oct 20 
02:07:24 2006
+++ llvm/lib/Transforms/Instrumentation/EmitFunctions.cpp   Sat Dec 30 
23:48:39 2006
@@ -87,7 +87,7 @@
   //std::cerrbegin();
   while (isa(InsertPos)) ++InsertPos;
 
-  std::vector GEPIndices(2, Constant::getNullValue(Type::IntTy));
+  std::vector GEPIndices(2, Constant::getNullValue(Type::Int32Ty));
   unsigned NumElements = 0;
   if (Array) {
 Args[2] = ConstantExpr::getGetElementPtr(Array, GEPIndices);
@@ -51,7 +51,7 @@
 // pass null.
 Args[2] = ConstantPointerNull::get(UIntPtr);
   }
-  Args[3] = ConstantInt::get(Type::UIntTy, NumElements);
+  Args[3] = ConstantInt::get(Type::Int32Ty, NumElements);
 
   Instruction *InitCall = new CallInst(InitFn, Args, "newargc", InsertPos);
 
@@ -75,16 +75,16 @@
 AI = MainFn->arg_begin();
 // If the program looked at argc, have it look at the return value of the
 // init call instead.
-if 

[llvm-commits] CVS: llvm/lib/Transforms/Utils/CodeExtractor.cpp InlineFunction.cpp LowerAllocations.cpp LowerInvoke.cpp LowerSwitch.cpp

[Reid Spencer]

Changes in directory llvm/lib/Transforms/Utils:

CodeExtractor.cpp updated: 1.43 -> 1.44
InlineFunction.cpp updated: 1.47 -> 1.48
LowerAllocations.cpp updated: 1.68 -> 1.69
LowerInvoke.cpp updated: 1.48 -> 1.49
LowerSwitch.cpp updated: 1.31 -> 1.32
---
Log message:

For PR950: http://llvm.org/PR950 :
This patch replaces signed integer types with signless ones:
1. [US]Byte -> Int8
2. [U]Short -> Int16
3. [U]Int   -> Int32
4. [U]Long  -> Int64.
5. Removal of isSigned, isUnsigned, getSignedVersion, getUnsignedVersion
   and other methods related to signedness. In a few places this warranted
   identifying the signedness information from other sources.



---
Diffs of the changes:  (+36 -38)

 CodeExtractor.cpp|   24 
 InlineFunction.cpp   |2 +-
 LowerAllocations.cpp |6 +++---
 LowerInvoke.cpp  |   38 +++---
 LowerSwitch.cpp  |4 +---
 5 files changed, 36 insertions(+), 38 deletions(-)


Index: llvm/lib/Transforms/Utils/CodeExtractor.cpp
diff -u llvm/lib/Transforms/Utils/CodeExtractor.cpp:1.43 
llvm/lib/Transforms/Utils/CodeExtractor.cpp:1.44
--- llvm/lib/Transforms/Utils/CodeExtractor.cpp:1.43Wed Dec  6 19:30:31 2006
+++ llvm/lib/Transforms/Utils/CodeExtractor.cpp Sat Dec 30 23:48:39 2006
@@ -252,7 +252,7 @@
   case 0:
   case 1: RetTy = Type::VoidTy; break;
   case 2: RetTy = Type::BoolTy; break;
-  default: RetTy = Type::UShortTy; break;
+  default: RetTy = Type::Int16Ty; break;
   }
 
   std::vector paramTy;
@@ -304,8 +304,8 @@
 Value *RewriteVal;
 if (AggregateArgs) {
   std::vector Indices;
-  Indices.push_back(Constant::getNullValue(Type::UIntTy));
-  Indices.push_back(ConstantInt::get(Type::UIntTy, i));
+  Indices.push_back(Constant::getNullValue(Type::Int32Ty));
+  Indices.push_back(ConstantInt::get(Type::Int32Ty, i));
   std::string GEPname = "gep_" + inputs[i]->getName();
   TerminatorInst *TI = newFunction->begin()->getTerminator();
   GetElementPtrInst *GEP = new GetElementPtrInst(AI, Indices, GEPname, TI);
@@ -391,8 +391,8 @@
 
 for (unsigned i = 0, e = inputs.size(); i != e; ++i) {
   std::vector Indices;
-  Indices.push_back(Constant::getNullValue(Type::UIntTy));
-  Indices.push_back(ConstantInt::get(Type::UIntTy, i));
+  Indices.push_back(Constant::getNullValue(Type::Int32Ty));
+  Indices.push_back(ConstantInt::get(Type::Int32Ty, i));
   GetElementPtrInst *GEP =
 new GetElementPtrInst(Struct, Indices,
   "gep_" + StructValues[i]->getName());
@@ -417,8 +417,8 @@
 Value *Output = 0;
 if (AggregateArgs) {
   std::vector Indices;
-  Indices.push_back(Constant::getNullValue(Type::UIntTy));
-  Indices.push_back(ConstantInt::get(Type::UIntTy, FirstOut + i));
+  Indices.push_back(Constant::getNullValue(Type::Int32Ty));
+  Indices.push_back(ConstantInt::get(Type::Int32Ty, FirstOut + i));
   GetElementPtrInst *GEP
 = new GetElementPtrInst(Struct, Indices,
 "gep_reload_" + outputs[i]->getName());
@@ -439,7 +439,7 @@
 
   // Now we can emit a switch statement using the call as a value.
   SwitchInst *TheSwitch =
-new SwitchInst(ConstantInt::getNullValue(Type::UShortTy),
+new SwitchInst(ConstantInt::getNullValue(Type::Int16Ty),
codeReplacer, 0, codeReplacer);
 
   // Since there may be multiple exits from the original region, make the new
@@ -473,14 +473,14 @@
 brVal = ConstantBool::get(!SuccNum);
 break;
   default:
-brVal = ConstantInt::get(Type::UShortTy, SuccNum);
+brVal = ConstantInt::get(Type::Int16Ty, SuccNum);
 break;
   }
 
   ReturnInst *NTRet = new ReturnInst(brVal, NewTarget);
 
   // Update the switch instruction.
-  TheSwitch->addCase(ConstantInt::get(Type::UShortTy, SuccNum),
+  TheSwitch->addCase(ConstantInt::get(Type::Int16Ty, SuccNum),
  OldTarget);
 
   // Restore values just before we exit
@@ -518,8 +518,8 @@
 if (DominatesDef) {
   if (AggregateArgs) {
 std::vector Indices;
-Indices.push_back(Constant::getNullValue(Type::UIntTy));
-Indices.push_back(ConstantInt::get(Type::UIntTy,FirstOut+out));
+Indices.push_back(Constant::getNullValue(Type::Int32Ty));
+
Indices.push_back(ConstantInt::get(Type::Int32Ty,FirstOut+out));
 GetElementPtrInst *GEP =
   new GetElementPtrInst(OAI, Indices,
 "gep_" + outputs[out]->getName(),


Index: llvm/lib/Transforms/Utils/InlineFunction.cpp
diff -u llvm/lib/Transforms/Utils/InlineFunction.cpp:1.47 
llvm/lib/Transforms/Utils/InlineFunction.cpp:1.48
--- llvm/lib/Transforms/Utils/InlineFunction.cpp:1.47   Wed Sep 13 14:23:57 2006
+++ llvm/lib/Transforms/Utils/Inl

[llvm-commits] CVS: llvm/examples/ModuleMaker/ModuleMaker.cpp

[Reid Spencer]

Changes in directory llvm/examples/ModuleMaker:

ModuleMaker.cpp updated: 1.11 -> 1.12
---
Log message:

For PR950: http://llvm.org/PR950 :
Convert signed integer types to signless.


---
Diffs of the changes:  (+3 -3)

 ModuleMaker.cpp |6 +++---
 1 files changed, 3 insertions(+), 3 deletions(-)


Index: llvm/examples/ModuleMaker/ModuleMaker.cpp
diff -u llvm/examples/ModuleMaker/ModuleMaker.cpp:1.11 
llvm/examples/ModuleMaker/ModuleMaker.cpp:1.12
--- llvm/examples/ModuleMaker/ModuleMaker.cpp:1.11  Wed Dec  6 19:30:30 2006
+++ llvm/examples/ModuleMaker/ModuleMaker.cpp   Sat Dec 30 23:50:28 2006
@@ -27,7 +27,7 @@
   Module *M = new Module("test");
 
   // Create the main function: first create the type 'int ()'
-  FunctionType *FT = FunctionType::get(Type::IntTy, std::vector(),
+  FunctionType *FT = FunctionType::get(Type::Int32Ty, std::vector(),
/*not vararg*/false);
 
   // By passing a module as the last parameter to the Function constructor,
@@ -39,8 +39,8 @@
   BasicBlock *BB = new BasicBlock("EntryBlock", F);
 
   // Get pointers to the constant integers...
-  Value *Two = ConstantInt::get(Type::IntTy, 2);
-  Value *Three = ConstantInt::get(Type::IntTy, 3);
+  Value *Two = ConstantInt::get(Type::Int32Ty, 2);
+  Value *Three = ConstantInt::get(Type::Int32Ty, 3);
 
   // Create the add instruction... does not insert...
   Instruction *Add = BinaryOperator::create(Instruction::Add, Two, Three,



___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

[llvm-commits] CVS: llvm/utils/TableGen/CodeGenIntrinsics.h

[Reid Spencer]

Changes in directory llvm/utils/TableGen:

CodeGenIntrinsics.h updated: 1.10 -> 1.11
---
Log message:

For PR950: http://llvm.org/PR950 :
Convert signed integer types to signless.


---
Diffs of the changes:  (+1 -1)

 CodeGenIntrinsics.h |2 +-
 1 files changed, 1 insertion(+), 1 deletion(-)


Index: llvm/utils/TableGen/CodeGenIntrinsics.h
diff -u llvm/utils/TableGen/CodeGenIntrinsics.h:1.10 
llvm/utils/TableGen/CodeGenIntrinsics.h:1.11
--- llvm/utils/TableGen/CodeGenIntrinsics.h:1.10Mon Mar 27 18:15:00 2006
+++ llvm/utils/TableGen/CodeGenIntrinsics.h Sat Dec 30 23:50:28 2006
@@ -31,7 +31,7 @@
 std::string TargetPrefix;  // Target prefix, e.g. "ppc" for t-s intrinsics.
 
 /// ArgTypes - The type primitive enum value for the return value and all
-/// of the arguments.  These are things like Type::UIntTyID.
+/// of the arguments.  These are things like Type::Int32TyID.
 std::vector ArgTypes;
 
 /// ArgVTs - The MVT::ValueType for each argument type.  Note that this 
list



___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

[llvm-commits] CVS: llvm/examples/ParallelJIT/ParallelJIT.cpp

[Reid Spencer]

Changes in directory llvm/examples/ParallelJIT:

ParallelJIT.cpp updated: 1.7 -> 1.8
---
Log message:

For PR950: http://llvm.org/PR950 :
Convert signed integer types to signless.


---
Diffs of the changes:  (+7 -7)

 ParallelJIT.cpp |   14 +++---
 1 files changed, 7 insertions(+), 7 deletions(-)


Index: llvm/examples/ParallelJIT/ParallelJIT.cpp
diff -u llvm/examples/ParallelJIT/ParallelJIT.cpp:1.7 
llvm/examples/ParallelJIT/ParallelJIT.cpp:1.8
--- llvm/examples/ParallelJIT/ParallelJIT.cpp:1.7   Sat Dec 23 00:05:40 2006
+++ llvm/examples/ParallelJIT/ParallelJIT.cpp   Sat Dec 30 23:50:28 2006
@@ -34,7 +34,7 @@
   // Create the add1 function entry and insert this entry into module M.  The
   // function will have a return type of "int" and take an argument of "int".
   // The '0' terminates the list of argument types.
-  Function *Add1F = M->getOrInsertFunction("add1", Type::IntTy, Type::IntTy,
+  Function *Add1F = M->getOrInsertFunction("add1", Type::Int32Ty, 
Type::Int32Ty,
(Type *)0);
 
   // Add a basic block to the function. As before, it automatically inserts
@@ -42,7 +42,7 @@
   BasicBlock *BB = new BasicBlock("EntryBlock", Add1F);
 
   // Get pointers to the constant `1'.
-  Value *One = ConstantInt::get(Type::IntTy, 1);
+  Value *One = ConstantInt::get(Type::Int32Ty, 1);
 
   // Get pointers to the integer argument of the add1 function...
   assert(Add1F->arg_begin() != Add1F->arg_end()); // Make sure there's an arg
@@ -63,15 +63,15 @@
 {
   // Create the fib function and insert it into module M.  This function is 
said
   // to return an int and take an int parameter.
-  Function *FibF = M->getOrInsertFunction("fib", Type::IntTy, Type::IntTy,
+  Function *FibF = M->getOrInsertFunction("fib", Type::Int32Ty, Type::Int32Ty,
   (Type *)0);
 
   // Add a basic block to the function.
   BasicBlock *BB = new BasicBlock("EntryBlock", FibF);
 
   // Get pointers to the constants.
-  Value *One = ConstantInt::get(Type::IntTy, 1);
-  Value *Two = ConstantInt::get(Type::IntTy, 2);
+  Value *One = ConstantInt::get(Type::Int32Ty, 1);
+  Value *Two = ConstantInt::get(Type::Int32Ty, 2);
 
   // Get pointer to the integer argument of the add1 function...
   Argument *ArgX = FibF->arg_begin();   // Get the arg.
@@ -221,12 +221,12 @@
 
   // Call the `foo' function with no arguments:
   std::vector Args(1);
-  Args[0].IntVal = p->value;
+  Args[0].Int32Val = p->value;
 
   synchronize.block(); // wait until other threads are at this point
   GenericValue gv = p->EE->runFunction(p->F, Args);
 
-  return (void*) intptr_t(gv.IntVal);
+  return (void*) intptr_t(gv.Int32Val);
 }
 
 int main()



___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

[llvm-commits] CVS: llvm/examples/HowToUseJIT/HowToUseJIT.cpp

[Reid Spencer]

Changes in directory llvm/examples/HowToUseJIT:

HowToUseJIT.cpp updated: 1.12 -> 1.13
---
Log message:

For PR950: http://llvm.org/PR950 :
Convert signed integer types to signless.


---
Diffs of the changes:  (+5 -5)

 HowToUseJIT.cpp |   10 +-
 1 files changed, 5 insertions(+), 5 deletions(-)


Index: llvm/examples/HowToUseJIT/HowToUseJIT.cpp
diff -u llvm/examples/HowToUseJIT/HowToUseJIT.cpp:1.12 
llvm/examples/HowToUseJIT/HowToUseJIT.cpp:1.13
--- llvm/examples/HowToUseJIT/HowToUseJIT.cpp:1.12  Fri Oct 20 02:07:23 2006
+++ llvm/examples/HowToUseJIT/HowToUseJIT.cpp   Sat Dec 30 23:50:28 2006
@@ -52,7 +52,7 @@
   // Create the add1 function entry and insert this entry into module M.  The
   // function will have a return type of "int" and take an argument of "int".
   // The '0' terminates the list of argument types.
-  Function *Add1F = M->getOrInsertFunction("add1", Type::IntTy, Type::IntTy,
+  Function *Add1F = M->getOrInsertFunction("add1", Type::Int32Ty, 
Type::Int32Ty,
(Type *)0);
 
   // Add a basic block to the function. As before, it automatically inserts
@@ -60,7 +60,7 @@
   BasicBlock *BB = new BasicBlock("EntryBlock", Add1F);
 
   // Get pointers to the constant `1'.
-  Value *One = ConstantInt::get(Type::IntTy, 1);
+  Value *One = ConstantInt::get(Type::Int32Ty, 1);
 
   // Get pointers to the integer argument of the add1 function...
   assert(Add1F->arg_begin() != Add1F->arg_end()); // Make sure there's an arg
@@ -78,13 +78,13 @@
 
   // Now we going to create function `foo', which returns an int and takes no
   // arguments.
-  Function *FooF = M->getOrInsertFunction("foo", Type::IntTy, (Type *)0);
+  Function *FooF = M->getOrInsertFunction("foo", Type::Int32Ty, (Type *)0);
 
   // Add a basic block to the FooF function.
   BB = new BasicBlock("EntryBlock", FooF);
 
   // Get pointers to the constant `10'.
-  Value *Ten = ConstantInt::get(Type::IntTy, 10);
+  Value *Ten = ConstantInt::get(Type::Int32Ty, 10);
 
   // Pass Ten to the call call:
   std::vector Params;
@@ -107,6 +107,6 @@
   GenericValue gv = EE->runFunction(FooF, noargs);
 
   // Import result of execution:
-  std::cout << "Result: " << gv.IntVal << "\n";
+  std::cout << "Result: " << gv.Int32Val << "\n";
   return 0;
 }



___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

[llvm-commits] CVS: llvm/projects/Stacker/lib/compiler/StackerCompiler.cpp

[Reid Spencer]

Changes in directory llvm/projects/Stacker/lib/compiler:

StackerCompiler.cpp updated: 1.28 -> 1.29
---
Log message:

For PR950: http://llvm.org/PR950 :
Convert signed integer types to signless.


---
Diffs of the changes:  (+44 -44)

 StackerCompiler.cpp |   88 ++--
 1 files changed, 44 insertions(+), 44 deletions(-)


Index: llvm/projects/Stacker/lib/compiler/StackerCompiler.cpp
diff -u llvm/projects/Stacker/lib/compiler/StackerCompiler.cpp:1.28 
llvm/projects/Stacker/lib/compiler/StackerCompiler.cpp:1.29
--- llvm/projects/Stacker/lib/compiler/StackerCompiler.cpp:1.28 Sat Dec 23 
00:05:41 2006
+++ llvm/projects/Stacker/lib/compiler/StackerCompiler.cpp  Sat Dec 30 
23:50:28 2006
@@ -122,7 +122,7 @@
 
 // Create a type to represent the stack. This is the same as the LLVM
 // Assembly type [ 256 x long ]
-stack_type = ArrayType::get( Type::LongTy, stack_size );
+stack_type = ArrayType::get( Type::Int64Ty, stack_size );
 
 // Create a global variable for the stack. Note the use of appending
 // linkage linkage so that multiple modules will make the stack larger.
@@ -141,10 +141,10 @@
 // of LinkOnce linkage. Only one copy of _index_ will be retained
 // after linking
 TheIndex = new GlobalVariable(
-/*type=*/Type::LongTy,
+/*type=*/Type::Int64Ty,
 /*isConstant=*/false,
 /*Linkage=*/GlobalValue::LinkOnceLinkage,
-/*initializer=*/ Constant::getNullValue(Type::LongTy),
+/*initializer=*/ Constant::getNullValue(Type::Int64Ty),
 /*name=*/"_index_",
 /*parent=*/TheModule
 );
@@ -155,9 +155,9 @@
 DefinitionType = FunctionType::get( Type::VoidTy, params, false );
 
 // Create a function for printf(3)
-params.push_back( PointerType::get( Type::SByteTy ) );
+params.push_back( PointerType::get( Type::Int8Ty ) );
 FunctionType* printf_type =
-FunctionType::get( Type::IntTy, params, true );
+FunctionType::get( Type::Int32Ty, params, true );
 ThePrintf = new Function(
 printf_type, GlobalValue::ExternalLinkage, "printf", TheModule);
 
@@ -167,7 +167,7 @@
 
 // Create a function for exit(3)
 params.clear();
-params.push_back( Type::IntTy );
+params.push_back( Type::Int32Ty );
 FunctionType* exit_type =
 FunctionType::get( Type::VoidTy, params, false );
 TheExit = new Function(
@@ -175,7 +175,7 @@
 
 Constant* str_format = ConstantArray::get("%s");
 StrFormat = new GlobalVariable(
-/*type=*/ArrayType::get( Type::SByteTy,  3 ),
+/*type=*/ArrayType::get( Type::Int8Ty,  3 ),
 /*isConstant=*/true,
 /*Linkage=*/GlobalValue::LinkOnceLinkage,
 /*initializer=*/str_format,
@@ -185,7 +185,7 @@
 
 Constant* in_str_format = ConstantArray::get(" %as");
 InStrFormat = new GlobalVariable(
-/*type=*/ArrayType::get( Type::SByteTy,  5 ),
+/*type=*/ArrayType::get( Type::Int8Ty,  5 ),
 /*isConstant=*/true,
 /*Linkage=*/GlobalValue::LinkOnceLinkage,
 /*initializer=*/in_str_format,
@@ -195,7 +195,7 @@
 
 Constant* num_format = ConstantArray::get("%d");
 NumFormat = new GlobalVariable(
-/*type=*/ArrayType::get( Type::SByteTy,  3 ),
+/*type=*/ArrayType::get( Type::Int8Ty,  3 ),
 /*isConstant=*/true,
 /*Linkage=*/GlobalValue::LinkOnceLinkage,
 /*initializer=*/num_format,
@@ -205,7 +205,7 @@
 
 Constant* in_num_format = ConstantArray::get(" %d");
 InNumFormat = new GlobalVariable(
-/*type=*/ArrayType::get( Type::SByteTy,  4 ),
+/*type=*/ArrayType::get( Type::Int8Ty,  4 ),
 /*isConstant=*/true,
 /*Linkage=*/GlobalValue::LinkOnceLinkage,
 /*initializer=*/in_num_format,
@@ -215,7 +215,7 @@
 
 Constant* chr_format = ConstantArray::get("%c");
 ChrFormat = new GlobalVariable(
-/*type=*/ArrayType::get( Type::SByteTy,  3 ),
+/*type=*/ArrayType::get( Type::Int8Ty,  3 ),
 /*isConstant=*/true,
 /*Linkage=*/GlobalValue::LinkOnceLinkage,
 /*initializer=*/chr_format,
@@ -225,7 +225,7 @@
 
 Constant* in_chr_format = ConstantArray::get(" %c");
 InChrFormat = new GlobalVariable(
-/*type=*/ArrayType::get( Type::SByteTy,  4 ),
+/*type=*/ArrayType::get( Type::Int8Ty,  4 ),
 /*isConstant=*/true,
 /*Linkage=*/GlobalValue::LinkOnceLinkage,
 /*initializer=*/in_chr_format,
@@ -234,12 +234,12 @@
 );
 
 // Get some constants so we aren't always creating them
-Zero = ConstantInt::get( Type::LongTy, 0 );
-One = ConstantInt::get( Type

[llvm-commits] CVS: llvm/examples/Fibonacci/fibonacci.cpp

[Reid Spencer]

Changes in directory llvm/examples/Fibonacci:

fibonacci.cpp updated: 1.13 -> 1.14
---
Log message:

For PR950: http://llvm.org/PR950 :
Convert signed integer types to signless.


---
Diffs of the changes:  (+5 -5)

 fibonacci.cpp |   10 +-
 1 files changed, 5 insertions(+), 5 deletions(-)


Index: llvm/examples/Fibonacci/fibonacci.cpp
diff -u llvm/examples/Fibonacci/fibonacci.cpp:1.13 
llvm/examples/Fibonacci/fibonacci.cpp:1.14
--- llvm/examples/Fibonacci/fibonacci.cpp:1.13  Sat Dec 23 00:05:40 2006
+++ llvm/examples/Fibonacci/fibonacci.cpp   Sat Dec 30 23:50:27 2006
@@ -38,15 +38,15 @@
 static Function *CreateFibFunction(Module *M) {
   // Create the fib function and insert it into module M.  This function is 
said
   // to return an int and take an int parameter.
-  Function *FibF = M->getOrInsertFunction("fib", Type::IntTy, Type::IntTy,
+  Function *FibF = M->getOrInsertFunction("fib", Type::Int32Ty, Type::Int32Ty,
   (Type *)0);
 
   // Add a basic block to the function.
   BasicBlock *BB = new BasicBlock("EntryBlock", FibF);
 
   // Get pointers to the constants.
-  Value *One = ConstantInt::get(Type::IntTy, 1);
-  Value *Two = ConstantInt::get(Type::IntTy, 2);
+  Value *One = ConstantInt::get(Type::Int32Ty, 1);
+  Value *Two = ConstantInt::get(Type::Int32Ty, 2);
 
   // Get pointer to the integer argument of the add1 function...
   Argument *ArgX = FibF->arg_begin();   // Get the arg.
@@ -111,10 +111,10 @@
 
   // Call the Fibonacci function with argument n:
   std::vector Args(1);
-  Args[0].IntVal = n;
+  Args[0].Int32Val = n;
   GenericValue GV = EE->runFunction(FibF, Args);
 
   // import result of execution
-  std::cout << "Result: " << GV.IntVal << "\n";
+  std::cout << "Result: " << GV.Int32Val << "\n";
   return 0;
 }



___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

[llvm-commits] CVS: llvm/lib/ExecutionEngine/JIT/JIT.cpp

[Reid Spencer]

Changes in directory llvm/lib/ExecutionEngine/JIT:

JIT.cpp updated: 1.84 -> 1.85
---
Log message:

For PR950: http://llvm.org/PR950 :
Convert signed integer types to signless ones.


---
Diffs of the changes:  (+24 -32)

 JIT.cpp |   56 
 1 files changed, 24 insertions(+), 32 deletions(-)


Index: llvm/lib/ExecutionEngine/JIT/JIT.cpp
diff -u llvm/lib/ExecutionEngine/JIT/JIT.cpp:1.84 
llvm/lib/ExecutionEngine/JIT/JIT.cpp:1.85
--- llvm/lib/ExecutionEngine/JIT/JIT.cpp:1.84   Sun Dec 17 15:04:02 2006
+++ llvm/lib/ExecutionEngine/JIT/JIT.cppSat Dec 30 23:51:36 2006
@@ -95,11 +95,11 @@
 
   // Handle some common cases first.  These cases correspond to common `main'
   // prototypes.
-  if (RetTy == Type::IntTy || RetTy == Type::UIntTy || RetTy == Type::VoidTy) {
+  if (RetTy == Type::Int32Ty || RetTy == Type::Int32Ty || RetTy == 
Type::VoidTy) {
 switch (ArgValues.size()) {
 case 3:
-  if ((FTy->getParamType(0) == Type::IntTy ||
-   FTy->getParamType(0) == Type::UIntTy) &&
+  if ((FTy->getParamType(0) == Type::Int32Ty ||
+   FTy->getParamType(0) == Type::Int32Ty) &&
   isa(FTy->getParamType(1)) &&
   isa(FTy->getParamType(2))) {
 int (*PF)(int, char **, const char **) =
@@ -107,30 +107,30 @@
 
 // Call the function.
 GenericValue rv;
-rv.IntVal = PF(ArgValues[0].IntVal, (char **)GVTOP(ArgValues[1]),
+rv.Int32Val = PF(ArgValues[0].Int32Val, (char **)GVTOP(ArgValues[1]),
(const char **)GVTOP(ArgValues[2]));
 return rv;
   }
   break;
 case 2:
-  if ((FTy->getParamType(0) == Type::IntTy ||
-   FTy->getParamType(0) == Type::UIntTy) &&
+  if ((FTy->getParamType(0) == Type::Int32Ty ||
+   FTy->getParamType(0) == Type::Int32Ty) &&
   isa(FTy->getParamType(1))) {
 int (*PF)(int, char **) = (int(*)(int, char **))(intptr_t)FPtr;
 
 // Call the function.
 GenericValue rv;
-rv.IntVal = PF(ArgValues[0].IntVal, (char **)GVTOP(ArgValues[1]));
+rv.Int32Val = PF(ArgValues[0].Int32Val, (char **)GVTOP(ArgValues[1]));
 return rv;
   }
   break;
 case 1:
   if (FTy->getNumParams() == 1 &&
-  (FTy->getParamType(0) == Type::IntTy ||
-   FTy->getParamType(0) == Type::UIntTy)) {
+  (FTy->getParamType(0) == Type::Int32Ty ||
+   FTy->getParamType(0) == Type::Int32Ty)) {
 GenericValue rv;
 int (*PF)(int) = (int(*)(int))(intptr_t)FPtr;
-rv.IntVal = PF(ArgValues[0].IntVal);
+rv.Int32Val = PF(ArgValues[0].Int32Val);
 return rv;
   }
   break;
@@ -145,22 +145,18 @@
 case Type::BoolTyID:
   rv.BoolVal = ((bool(*)())(intptr_t)FPtr)();
   return rv;
-case Type::SByteTyID:
-case Type::UByteTyID:
-  rv.SByteVal = ((char(*)())(intptr_t)FPtr)();
+case Type::Int8TyID:
+  rv.Int8Val = ((char(*)())(intptr_t)FPtr)();
   return rv;
-case Type::ShortTyID:
-case Type::UShortTyID:
-  rv.ShortVal = ((short(*)())(intptr_t)FPtr)();
+case Type::Int16TyID:
+  rv.Int16Val = ((short(*)())(intptr_t)FPtr)();
   return rv;
 case Type::VoidTyID:
-case Type::IntTyID:
-case Type::UIntTyID:
-  rv.IntVal = ((int(*)())(intptr_t)FPtr)();
+case Type::Int32TyID:
+  rv.Int32Val = ((int(*)())(intptr_t)FPtr)();
   return rv;
-case Type::LongTyID:
-case Type::ULongTyID:
-  rv.LongVal = ((int64_t(*)())(intptr_t)FPtr)();
+case Type::Int64TyID:
+  rv.Int64Val = ((int64_t(*)())(intptr_t)FPtr)();
   return rv;
 case Type::FloatTyID:
   rv.FloatVal = ((float(*)())(intptr_t)FPtr)();
@@ -196,22 +192,18 @@
 switch (ArgTy->getTypeID()) {
 default: assert(0 && "Unknown argument type for function call!");
 case Type::BoolTyID:   C = ConstantBool::get(AV.BoolVal); break;
-case Type::SByteTyID:  C = ConstantInt::get(ArgTy, AV.SByteVal);  break;
-case Type::UByteTyID:  C = ConstantInt::get(ArgTy, AV.UByteVal);  break;
-case Type::ShortTyID:  C = ConstantInt::get(ArgTy, AV.ShortVal);  break;
-case Type::UShortTyID: C = ConstantInt::get(ArgTy, AV.UShortVal); break;
-case Type::IntTyID:C = ConstantInt::get(ArgTy, AV.IntVal);break;
-case Type::UIntTyID:   C = ConstantInt::get(ArgTy, AV.UIntVal);   break;
-case Type::LongTyID:   C = ConstantInt::get(ArgTy, AV.LongVal);   break;
-case Type::ULongTyID:  C = ConstantInt::get(ArgTy, AV.ULongVal);  break;
+case Type::Int8TyID:   C = ConstantInt::get(ArgTy, AV.Int8Val);  break;
+case Type::Int16TyID:  C = ConstantInt::get(ArgTy, AV.Int16Val);  break;
+case Type::Int32TyID:  C = ConstantInt::get(ArgTy, AV.Int32Val);break;
+case Type::Int64TyID:  C = ConstantInt::get(ArgTy, AV.Int64Val);   break;
 case Type::FloatTyID:  C = ConstantFP ::get(ArgTy, AV.FloatVal);  break;
 case Type::DoubleTyID: C

[llvm-commits] CVS: llvm/lib/ExecutionEngine/Interpreter/Execution.cpp ExternalFunctions.cpp

[Reid Spencer]

Changes in directory llvm/lib/ExecutionEngine/Interpreter:

Execution.cpp updated: 1.154 -> 1.155
ExternalFunctions.cpp updated: 1.92 -> 1.93
---
Log message:

For PR950: http://llvm.org/PR950 :
Convert signed integer types to signless ones.


---
Diffs of the changes:  (+253 -369)

 Execution.cpp |  526 +++---
 ExternalFunctions.cpp |   96 -
 2 files changed, 253 insertions(+), 369 deletions(-)


Index: llvm/lib/ExecutionEngine/Interpreter/Execution.cpp
diff -u llvm/lib/ExecutionEngine/Interpreter/Execution.cpp:1.154 
llvm/lib/ExecutionEngine/Interpreter/Execution.cpp:1.155
--- llvm/lib/ExecutionEngine/Interpreter/Execution.cpp:1.154Sat Dec 23 
00:05:41 2006
+++ llvm/lib/ExecutionEngine/Interpreter/Execution.cpp  Sat Dec 30 23:51:36 2006
@@ -194,14 +194,10 @@
const Type *Ty) {
   GenericValue Dest;
   switch (Ty->getTypeID()) {
-IMPLEMENT_BINARY_OPERATOR(+, UByte);
-IMPLEMENT_BINARY_OPERATOR(+, SByte);
-IMPLEMENT_BINARY_OPERATOR(+, UShort);
-IMPLEMENT_BINARY_OPERATOR(+, Short);
-IMPLEMENT_BINARY_OPERATOR(+, UInt);
-IMPLEMENT_BINARY_OPERATOR(+, Int);
-IMPLEMENT_BINARY_OPERATOR(+, ULong);
-IMPLEMENT_BINARY_OPERATOR(+, Long);
+IMPLEMENT_BINARY_OPERATOR(+, Int8);
+IMPLEMENT_BINARY_OPERATOR(+, Int16);
+IMPLEMENT_BINARY_OPERATOR(+, Int32);
+IMPLEMENT_BINARY_OPERATOR(+, Int64);
 IMPLEMENT_BINARY_OPERATOR(+, Float);
 IMPLEMENT_BINARY_OPERATOR(+, Double);
   default:
@@ -215,14 +211,10 @@
const Type *Ty) {
   GenericValue Dest;
   switch (Ty->getTypeID()) {
-IMPLEMENT_BINARY_OPERATOR(-, UByte);
-IMPLEMENT_BINARY_OPERATOR(-, SByte);
-IMPLEMENT_BINARY_OPERATOR(-, UShort);
-IMPLEMENT_BINARY_OPERATOR(-, Short);
-IMPLEMENT_BINARY_OPERATOR(-, UInt);
-IMPLEMENT_BINARY_OPERATOR(-, Int);
-IMPLEMENT_BINARY_OPERATOR(-, ULong);
-IMPLEMENT_BINARY_OPERATOR(-, Long);
+IMPLEMENT_BINARY_OPERATOR(-, Int8);
+IMPLEMENT_BINARY_OPERATOR(-, Int16);
+IMPLEMENT_BINARY_OPERATOR(-, Int32);
+IMPLEMENT_BINARY_OPERATOR(-, Int64);
 IMPLEMENT_BINARY_OPERATOR(-, Float);
 IMPLEMENT_BINARY_OPERATOR(-, Double);
   default:
@@ -236,14 +228,10 @@
const Type *Ty) {
   GenericValue Dest;
   switch (Ty->getTypeID()) {
-IMPLEMENT_BINARY_OPERATOR(*, UByte);
-IMPLEMENT_BINARY_OPERATOR(*, SByte);
-IMPLEMENT_BINARY_OPERATOR(*, UShort);
-IMPLEMENT_BINARY_OPERATOR(*, Short);
-IMPLEMENT_BINARY_OPERATOR(*, UInt);
-IMPLEMENT_BINARY_OPERATOR(*, Int);
-IMPLEMENT_BINARY_OPERATOR(*, ULong);
-IMPLEMENT_BINARY_OPERATOR(*, Long);
+IMPLEMENT_BINARY_OPERATOR(*, Int8);
+IMPLEMENT_BINARY_OPERATOR(*, Int16);
+IMPLEMENT_BINARY_OPERATOR(*, Int32);
+IMPLEMENT_BINARY_OPERATOR(*, Int64);
 IMPLEMENT_BINARY_OPERATOR(*, Float);
 IMPLEMENT_BINARY_OPERATOR(*, Double);
   default:
@@ -253,17 +241,18 @@
   return Dest;
 }
 
-#define IMPLEMENT_SIGNLESS_BINOP(OP, TY1, TY2) \
-   case Type::TY2##TyID: IMPLEMENT_BINARY_OPERATOR(OP, TY1)
+#define IMPLEMENT_SIGNLESS_BINOP(OP, TY, CAST) \
+   case Type::TY##TyID: Dest.TY##Val = \
+((CAST)Src1.TY##Val) OP ((CAST)Src2.TY##Val); break
 
 static GenericValue executeUDivInst(GenericValue Src1, GenericValue Src2,
const Type *Ty) {
   GenericValue Dest;
   switch (Ty->getTypeID()) {
-IMPLEMENT_SIGNLESS_BINOP(/, UByte,  SByte);
-IMPLEMENT_SIGNLESS_BINOP(/, UShort, Short);
-IMPLEMENT_SIGNLESS_BINOP(/, UInt,   Int);
-IMPLEMENT_SIGNLESS_BINOP(/, ULong,  Long);
+IMPLEMENT_SIGNLESS_BINOP(/, Int8,  uint8_t);
+IMPLEMENT_SIGNLESS_BINOP(/, Int16, uint16_t);
+IMPLEMENT_SIGNLESS_BINOP(/, Int32, uint32_t);
+IMPLEMENT_SIGNLESS_BINOP(/, Int64, uint64_t);
   default:
 cerr << "Unhandled type for UDiv instruction: " << *Ty << "\n";
 abort();
@@ -275,10 +264,10 @@
const Type *Ty) {
   GenericValue Dest;
   switch (Ty->getTypeID()) {
-IMPLEMENT_SIGNLESS_BINOP(/, SByte, UByte);
-IMPLEMENT_SIGNLESS_BINOP(/, Short, UShort);
-IMPLEMENT_SIGNLESS_BINOP(/, Int,   UInt);
-IMPLEMENT_SIGNLESS_BINOP(/, Long,  ULong);
+IMPLEMENT_SIGNLESS_BINOP(/, Int8,  int8_t);
+IMPLEMENT_SIGNLESS_BINOP(/, Int16, int16_t);
+IMPLEMENT_SIGNLESS_BINOP(/, Int32, int32_t);
+IMPLEMENT_SIGNLESS_BINOP(/, Int64, int64_t);
   default:
 cerr << "Unhandled type for SDiv instruction: " << *Ty << "\n";
 abort();
@@ -303,10 +292,10 @@
const Type *Ty) {
   GenericValue Dest;
   switch (Ty->getTypeID()) {
-IMPLEMENT_SIGNLESS_BINOP(%, UByte,  SByte);
-IMPLEMENT_SIGNLESS_BINOP(%, UShort, Short);
-IMPLEMENT_SIGNLESS_BINOP(%, UInt,   Int);
-IMPLEMENT_SIGNLESS_BINOP(%, ULong,  Long);
+IMPLEMENT_SIGNLESS_BINOP(%, Int8,  uint8_t);
+IMPLEMENT_SIGNLESS_BINOP(%, Int16, uint16_

[llvm-commits] CVS: llvm/lib/ExecutionEngine/ExecutionEngine.cpp

[Reid Spencer]

Changes in directory llvm/lib/ExecutionEngine:

ExecutionEngine.cpp updated: 1.93 -> 1.94
---
Log message:

For PR950: http://llvm.org/PR950 :
Convert signed integer types to signless ones.


---
Diffs of the changes:  (+69 -97)

 ExecutionEngine.cpp |  166 +---
 1 files changed, 69 insertions(+), 97 deletions(-)


Index: llvm/lib/ExecutionEngine/ExecutionEngine.cpp
diff -u llvm/lib/ExecutionEngine/ExecutionEngine.cpp:1.93 
llvm/lib/ExecutionEngine/ExecutionEngine.cpp:1.94
--- llvm/lib/ExecutionEngine/ExecutionEngine.cpp:1.93   Tue Dec 19 16:43:32 2006
+++ llvm/lib/ExecutionEngine/ExecutionEngine.cppSat Dec 30 23:51:35 2006
@@ -157,7 +157,7 @@
   char *Result = new char[(InputArgv.size()+1)*PtrSize];
 
   DOUT << "ARGV = " << (void*)Result << "\n";
-  const Type *SBytePtr = PointerType::get(Type::SByteTy);
+  const Type *SBytePtr = PointerType::get(Type::Int8Ty);
 
   for (unsigned i = 0; i != InputArgv.size(); ++i) {
 unsigned Size = InputArgv[i].size()+1;
@@ -228,7 +228,7 @@
const char * const * envp) {
   std::vector GVArgs;
   GenericValue GVArgc;
-  GVArgc.IntVal = argv.size();
+  GVArgc.Int32Val = argv.size();
   unsigned NumArgs = Fn->getFunctionType()->getNumParams();
   if (NumArgs) {
 GVArgs.push_back(GVArgc); // Arg #0 = argc.
@@ -244,7 +244,7 @@
   }
 }
   }
-  return runFunction(Fn, GVArgs).IntVal;
+  return runFunction(Fn, GVArgs).Int32Val;
 }
 
 /// If possible, create a JIT, unless the caller specifically requests an
@@ -317,9 +317,9 @@
 TD->getIndexedOffset(CE->getOperand(0)->getType(), Indexes);
 
   if (getTargetData()->getPointerSize() == 4)
-Result.IntVal += Offset;
+Result.Int32Val += Offset;
   else
-Result.LongVal += Offset;
+Result.Int64Val += Offset;
   return Result;
 }
 case Instruction::Trunc:
@@ -352,14 +352,10 @@
   GenericValue GV = getConstantValue(Op);
   switch (Op->getType()->getTypeID()) {
 case Type::BoolTyID:return PTOGV((void*)(uintptr_t)GV.BoolVal);
-case Type::SByteTyID:   return PTOGV((void*)( intptr_t)GV.SByteVal);
-case Type::UByteTyID:   return PTOGV((void*)(uintptr_t)GV.UByteVal);
-case Type::ShortTyID:   return PTOGV((void*)( intptr_t)GV.ShortVal);
-case Type::UShortTyID:  return PTOGV((void*)(uintptr_t)GV.UShortVal);
-case Type::IntTyID: return PTOGV((void*)( intptr_t)GV.IntVal);
-case Type::UIntTyID:return PTOGV((void*)(uintptr_t)GV.UIntVal);
-case Type::LongTyID:return PTOGV((void*)( intptr_t)GV.LongVal);
-case Type::ULongTyID:   return PTOGV((void*)(uintptr_t)GV.ULongVal);
+case Type::Int8TyID:   return PTOGV((void*)(uintptr_t)GV.Int8Val);
+case Type::Int16TyID:  return PTOGV((void*)(uintptr_t)GV.Int16Val);
+case Type::Int32TyID:return PTOGV((void*)(uintptr_t)GV.Int32Val);
+case Type::Int64TyID:   return PTOGV((void*)(uintptr_t)GV.Int64Val);
 default: assert(0 && "Unknown integral type!");
   }
   break;
@@ -367,25 +363,21 @@
 case Instruction::Add:
   switch (CE->getOperand(0)->getType()->getTypeID()) {
   default: assert(0 && "Bad add type!"); abort();
-  case Type::LongTyID:
-  case Type::ULongTyID:
-Result.LongVal = getConstantValue(CE->getOperand(0)).LongVal +
- getConstantValue(CE->getOperand(1)).LongVal;
+  case Type::Int64TyID:
+Result.Int64Val = getConstantValue(CE->getOperand(0)).Int64Val +
+ getConstantValue(CE->getOperand(1)).Int64Val;
 break;
-  case Type::IntTyID:
-  case Type::UIntTyID:
-Result.IntVal = getConstantValue(CE->getOperand(0)).IntVal +
-getConstantValue(CE->getOperand(1)).IntVal;
+  case Type::Int32TyID:
+Result.Int32Val = getConstantValue(CE->getOperand(0)).Int32Val +
+getConstantValue(CE->getOperand(1)).Int32Val;
 break;
-  case Type::ShortTyID:
-  case Type::UShortTyID:
-Result.ShortVal = getConstantValue(CE->getOperand(0)).ShortVal +
-  getConstantValue(CE->getOperand(1)).ShortVal;
+  case Type::Int16TyID:
+Result.Int16Val = getConstantValue(CE->getOperand(0)).Int16Val +
+  getConstantValue(CE->getOperand(1)).Int16Val;
 break;
-  case Type::SByteTyID:
-  case Type::UByteTyID:
-Result.SByteVal = getConstantValue(CE->getOperand(0)).SByteVal +
-  getConstantValue(CE->getOperand(1)).SByteVal;
+  case Type::Int8TyID:
+Result.Int8Val = getConstantValue(CE->getOperand(0)).Int8Val +
+  getConstantValue(CE->getOperand(1)).Int8Val;
 break;
   case Type::FloatTyID:
 Result.FloatVal = getConstantValue(CE->getOperand(0)).FloatVal +
@@ -407,17 +399,13 @@
   switch (C->g

[llvm-commits] CVS: llvm/lib/CodeGen/AsmPrinter.cpp IntrinsicLowering.cpp MachOWriter.cpp MachineDebugInfo.cpp

[Reid Spencer]

Changes in directory llvm/lib/CodeGen:

AsmPrinter.cpp updated: 1.128 -> 1.129
IntrinsicLowering.cpp updated: 1.54 -> 1.55
MachOWriter.cpp updated: 1.9 -> 1.10
MachineDebugInfo.cpp updated: 1.64 -> 1.65
---
Log message:

For PR950: http://llvm.org/PR950 :
Three changes:
1. Convert signed integer types to signless versions.
2. Implement the @sext and @zext parameter attributes. Previously the
   type of an function parameter was used to determine whether it should
   be sign extended or zero extended before the call. This information is
   now communicated via the function type's parameter attributes. 
3. The interface to LowerCallTo had to be changed in order to accommodate
   the parameter attribute information. Although it would have been
   convenient to pass in the FunctionType itself, there isn't always one
   present in the caller. Consequently, a signedness indication for the
   result type and for each parameter was provided for in the interface
   to this method. All implementations were changed to make the adjustment
   necessary.


---
Diffs of the changes:  (+50 -54)

 AsmPrinter.cpp|   10 
 IntrinsicLowering.cpp |   60 +-
 MachOWriter.cpp   |   12 +++---
 MachineDebugInfo.cpp  |   22 +-
 4 files changed, 50 insertions(+), 54 deletions(-)


Index: llvm/lib/CodeGen/AsmPrinter.cpp
diff -u llvm/lib/CodeGen/AsmPrinter.cpp:1.128 
llvm/lib/CodeGen/AsmPrinter.cpp:1.129
--- llvm/lib/CodeGen/AsmPrinter.cpp:1.128   Thu Dec 21 13:04:23 2006
+++ llvm/lib/CodeGen/AsmPrinter.cpp Sat Dec 30 23:55:36 2006
@@ -607,7 +607,7 @@
 << "\t" << TAI->getCommentString() << " float " << Val << "\n";
   return;
 }
-  } else if (CV->getType() == Type::ULongTy || CV->getType() == Type::LongTy) {
+  } else if (CV->getType() == Type::Int64Ty) {
 if (const ConstantInt *CI = dyn_cast(CV)) {
   uint64_t Val = CI->getZExtValue();
 
@@ -918,10 +918,10 @@
   const TargetData *TD = TM.getTargetData();
   switch (type->getTypeID()) {
   case Type::BoolTyID:
-  case Type::UByteTyID: case Type::SByteTyID:
+  case Type::Int8TyID:
 O << TAI->getData8bitsDirective();
 break;
-  case Type::UShortTyID: case Type::ShortTyID:
+  case Type::Int16TyID: 
 O << TAI->getData16bitsDirective();
 break;
   case Type::PointerTyID:
@@ -932,10 +932,10 @@
   break;
 }
 //Fall through for pointer size == int size
-  case Type::UIntTyID: case Type::IntTyID:
+  case Type::Int32TyID: 
 O << TAI->getData32bitsDirective();
 break;
-  case Type::ULongTyID: case Type::LongTyID:
+  case Type::Int64TyID:
 assert(TAI->getData64bitsDirective() &&
"Target cannot handle 64-bit constant exprs!");
 O << TAI->getData64bitsDirective();


Index: llvm/lib/CodeGen/IntrinsicLowering.cpp
diff -u llvm/lib/CodeGen/IntrinsicLowering.cpp:1.54 
llvm/lib/CodeGen/IntrinsicLowering.cpp:1.55
--- llvm/lib/CodeGen/IntrinsicLowering.cpp:1.54 Sat Dec 23 00:05:40 2006
+++ llvm/lib/CodeGen/IntrinsicLowering.cpp  Sat Dec 30 23:55:36 2006
@@ -95,7 +95,7 @@
   default: break;
   case Intrinsic::setjmp:
 EnsureFunctionExists(M, "setjmp", I->arg_begin(), I->arg_end(),
- Type::IntTy);
+ Type::Int32Ty);
 break;
   case Intrinsic::longjmp:
 EnsureFunctionExists(M, "longjmp", I->arg_begin(), I->arg_end(),
@@ -117,9 +117,9 @@
 break;
   case Intrinsic::memset_i32:
   case Intrinsic::memset_i64:
-M.getOrInsertFunction("memset", PointerType::get(Type::SByteTy),
-  PointerType::get(Type::SByteTy),
-  Type::IntTy, (--(--I->arg_end()))->getType(),
+M.getOrInsertFunction("memset", PointerType::get(Type::Int8Ty),
+  PointerType::get(Type::Int8Ty),
+  Type::Int32Ty, (--(--I->arg_end()))->getType(),
   (Type *)0);
 break;
   case Intrinsic::isunordered_f32:
@@ -150,26 +150,26 @@
   default: assert(0 && "Unhandled type size of value to byteswap!");
   case 16: {
 Value *Tmp1 = new ShiftInst(Instruction::Shl, V,
-
ConstantInt::get(Type::UByteTy,8),"bswap.2",IP);
+ConstantInt::get(Type::Int8Ty,8),"bswap.2",IP);
 Value *Tmp2 = new ShiftInst(Instruction::LShr, V,
-
ConstantInt::get(Type::UByteTy,8),"bswap.1",IP);
+ConstantInt::get(Type::Int8Ty,8),"bswap.1",IP);
 V = BinaryOperator::createOr(Tmp1, Tmp2, "bswap.i16", IP);
 break;
   }
   case 32: {
 Value *Tmp4 = new ShiftInst(Instruction::Shl, V,
-  ConstantInt::get(Type::UByteTy,24),"bswap.4", 
IP);
+  ConstantInt::get(Type::Int8Ty,24),"bswap.4", IP);
 Value *Tmp3 = new ShiftInst(Instruction::Shl, V,
-  

[llvm-commits] CVS: llvm/lib/Target/ARM/ARMISelDAGToDAG.cpp

[Reid Spencer]

Changes in directory llvm/lib/Target/ARM:

ARMISelDAGToDAG.cpp updated: 1.94 -> 1.95
---
Log message:

For PR950: http://llvm.org/PR950 :
Three changes:
1. Convert signed integer types to signless versions.
2. Implement the @sext and @zext parameter attributes. Previously the
   type of an function parameter was used to determine whether it should
   be sign extended or zero extended before the call. This information is
   now communicated via the function type's parameter attributes. 
3. The interface to LowerCallTo had to be changed in order to accommodate
   the parameter attribute information. Although it would have been
   convenient to pass in the FunctionType itself, there isn't always one
   present in the caller. Consequently, a signedness indication for the
   result type and for each parameter was provided for in the interface
   to this method. All implementations were changed to make the adjustment
   necessary.


---
Diffs of the changes:  (+1 -1)

 ARMISelDAGToDAG.cpp |2 +-
 1 files changed, 1 insertion(+), 1 deletion(-)


Index: llvm/lib/Target/ARM/ARMISelDAGToDAG.cpp
diff -u llvm/lib/Target/ARM/ARMISelDAGToDAG.cpp:1.94 
llvm/lib/Target/ARM/ARMISelDAGToDAG.cpp:1.95
--- llvm/lib/Target/ARM/ARMISelDAGToDAG.cpp:1.94Fri Dec 29 08:28:12 2006
+++ llvm/lib/Target/ARM/ARMISelDAGToDAG.cpp Sat Dec 30 23:55:36 2006
@@ -915,7 +915,7 @@
 SDOperand C = CurDAG->getTargetConstant(~val,  MVT::i32);
 n   = CurDAG->getTargetNode(ARM::MVN,  MVT::i32, C, Z, Z);
  } else {
-Constant*C = ConstantInt::get(Type::UIntTy, val);
+Constant*C = ConstantInt::get(Type::Int32Ty, val);
 int  alignment = 2;
 SDOperand Addr = CurDAG->getTargetConstantPool(C, MVT::i32, alignment);
 n  = CurDAG->getTargetNode(ARM::LDR,  MVT::i32, Addr, Z);



___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

[llvm-commits] CVS: llvm/lib/Target/Sparc/SparcISelDAGToDAG.cpp

[Reid Spencer]

Changes in directory llvm/lib/Target/Sparc:

SparcISelDAGToDAG.cpp updated: 1.116 -> 1.117
---
Log message:

For PR950: http://llvm.org/PR950 :
Three changes:
1. Convert signed integer types to signless versions.
2. Implement the @sext and @zext parameter attributes. Previously the
   type of an function parameter was used to determine whether it should
   be sign extended or zero extended before the call. This information is
   now communicated via the function type's parameter attributes. 
3. The interface to LowerCallTo had to be changed in order to accommodate
   the parameter attribute information. Although it would have been
   convenient to pass in the FunctionType itself, there isn't always one
   present in the caller. Consequently, a signedness indication for the
   result type and for each parameter was provided for in the interface
   to this method. All implementations were changed to make the adjustment
   necessary.


---
Diffs of the changes:  (+21 -19)

 SparcISelDAGToDAG.cpp |   40 +---
 1 files changed, 21 insertions(+), 19 deletions(-)


Index: llvm/lib/Target/Sparc/SparcISelDAGToDAG.cpp
diff -u llvm/lib/Target/Sparc/SparcISelDAGToDAG.cpp:1.116 
llvm/lib/Target/Sparc/SparcISelDAGToDAG.cpp:1.117
--- llvm/lib/Target/Sparc/SparcISelDAGToDAG.cpp:1.116   Thu Dec  7 16:21:48 2006
+++ llvm/lib/Target/Sparc/SparcISelDAGToDAG.cpp Sat Dec 30 23:55:36 2006
@@ -117,10 +117,9 @@
 virtual std::vector
   LowerArguments(Function &F, SelectionDAG &DAG);
 virtual std::pair
-  LowerCallTo(SDOperand Chain, const Type *RetTy, bool isVarArg,
-  unsigned CC,
-  bool isTailCall, SDOperand Callee, ArgListTy &Args,
-  SelectionDAG &DAG);
+  LowerCallTo(SDOperand Chain, const Type *RetTy, bool RetTyIsSigned, 
+  bool isVarArg, unsigned CC, bool isTailCall, SDOperand 
Callee,
+  ArgListTy &Args, SelectionDAG &DAG);
 virtual MachineBasicBlock *InsertAtEndOfBasicBlock(MachineInstr *MI,
MachineBasicBlock *MBB);
 
@@ -318,8 +317,7 @@
 MF.addLiveIn(*CurArgReg++, VReg);
 SDOperand Arg = DAG.getCopyFromReg(Root, VReg, MVT::i32);
 if (ObjectVT != MVT::i32) {
-  unsigned AssertOp = I->getType()->isSigned() ? ISD::AssertSext 
-   : ISD::AssertZext;
+  unsigned AssertOp = ISD::AssertSext;
   Arg = DAG.getNode(AssertOp, MVT::i32, Arg, 
 DAG.getValueType(ObjectVT));
   Arg = DAG.getNode(ISD::TRUNCATE, ObjectVT, Arg);
@@ -332,8 +330,7 @@
 if (ObjectVT == MVT::i32) {
   Load = DAG.getLoad(MVT::i32, Root, FIPtr, NULL, 0);
 } else {
-  ISD::LoadExtType LoadOp =
-I->getType()->isSigned() ? ISD::SEXTLOAD : ISD::ZEXTLOAD;
+  ISD::LoadExtType LoadOp = ISD::SEXTLOAD;
 
   // Sparc is big endian, so add an offset based on the ObjectVT.
   unsigned Offset = 4-std::max(1U, MVT::getSizeInBits(ObjectVT)/8);
@@ -472,13 +469,13 @@
 
 std::pair
 SparcTargetLowering::LowerCallTo(SDOperand Chain, const Type *RetTy,
- bool isVarArg, unsigned CC,
+ bool RetTyIsSigned, bool isVarArg, unsigned 
CC,
  bool isTailCall, SDOperand Callee, 
  ArgListTy &Args, SelectionDAG &DAG) {
   // Count the size of the outgoing arguments.
   unsigned ArgsSize = 0;
   for (unsigned i = 0, e = Args.size(); i != e; ++i) {
-switch (getValueType(Args[i].second)) {
+switch (getValueType(Args[i].Ty)) {
 default: assert(0 && "Unknown value type!");
 case MVT::i1:
 case MVT::i8:
@@ -508,7 +505,7 @@
   std::vector RegValuesToPass;
   unsigned ArgOffset = 68;
   for (unsigned i = 0, e = Args.size(); i != e; ++i) {
-SDOperand Val = Args[i].first;
+SDOperand Val = Args[i].Node;
 MVT::ValueType ObjectVT = Val.getValueType();
 SDOperand ValToStore(0, 0);
 unsigned ObjSize;
@@ -516,14 +513,15 @@
 default: assert(0 && "Unhandled argument type!");
 case MVT::i1:
 case MVT::i8:
-case MVT::i16:
+case MVT::i16: {
   // Promote the integer to 32-bits.  If the input type is signed, use a
   // sign extend, otherwise use a zero extend.
-  if (Args[i].second->isSigned())
-Val = DAG.getNode(ISD::SIGN_EXTEND, MVT::i32, Val);
-  else
-Val = DAG.getNode(ISD::ZERO_EXTEND, MVT::i32, Val);
+  ISD::NodeType ExtendKind = ISD::ZERO_EXTEND;
+  if (Args[i].isSigned)
+ExtendKind = ISD::SIGN_EXTEND;
+  Val = DAG.getNode(ExtendKind, MVT::i32, Val);
   // FALL THROUGH
+}
 case MVT::i32:
   ObjSize = 4;
 
@@ -629,15 +627,19 @@
 default: assert(0 && "Unknown value type to return!");
 case MVT::i1:
 case MVT::i8:
-case MVT::i16:
+case MVT::

[llvm-commits] CVS: llvm/lib/Target/CBackend/Writer.cpp

[Reid Spencer]

Changes in directory llvm/lib/Target/CBackend:

Writer.cpp updated: 1.301 -> 1.302
---
Log message:

For PR950: http://llvm.org/PR950 :
Three changes:
1. Convert signed integer types to signless versions.
2. Implement the @sext and @zext parameter attributes. Previously the
   type of an function parameter was used to determine whether it should
   be sign extended or zero extended before the call. This information is
   now communicated via the function type's parameter attributes. 
3. The interface to LowerCallTo had to be changed in order to accommodate
   the parameter attribute information. Although it would have been
   convenient to pass in the FunctionType itself, there isn't always one
   present in the caller. Consequently, a signedness indication for the
   result type and for each parameter was provided for in the interface
   to this method. All implementations were changed to make the adjustment
   necessary.


---
Diffs of the changes:  (+16 -56)

 Writer.cpp |   72 +
 1 files changed, 16 insertions(+), 56 deletions(-)


Index: llvm/lib/Target/CBackend/Writer.cpp
diff -u llvm/lib/Target/CBackend/Writer.cpp:1.301 
llvm/lib/Target/CBackend/Writer.cpp:1.302
--- llvm/lib/Target/CBackend/Writer.cpp:1.301   Sat Dec 23 00:05:41 2006
+++ llvm/lib/Target/CBackend/Writer.cpp Sat Dec 30 23:55:36 2006
@@ -365,17 +365,13 @@
   switch (Ty->getTypeID()) {
   case Type::VoidTyID:   return Out << "void "   << NameSoFar;
   case Type::BoolTyID:   return Out << "bool "   << NameSoFar;
-  case Type::UByteTyID:  
-  case Type::SByteTyID:
+  case Type::Int8TyID:
 return Out << (isSigned?"signed":"unsigned") << " char " << NameSoFar;
-  case Type::UShortTyID: 
-  case Type::ShortTyID:  
+  case Type::Int16TyID:  
 return Out << (isSigned?"signed":"unsigned") << " short " << NameSoFar;
-  case Type::UIntTyID:   
-  case Type::IntTyID:
+  case Type::Int32TyID:
 return Out << (isSigned?"signed":"unsigned") << " int " << NameSoFar;
-  case Type::ULongTyID:  
-  case Type::LongTyID:   
+  case Type::Int64TyID:   
 return Out << (isSigned?"signed":"unsigned") << " long long " << NameSoFar;
   case Type::FloatTyID:  return Out << "float "  << NameSoFar;
   case Type::DoubleTyID: return Out << "double " << NameSoFar;
@@ -488,7 +484,7 @@
   // ubytes or an array of sbytes with positive values.
   //
   const Type *ETy = CPA->getType()->getElementType();
-  bool isString = (ETy == Type::SByteTy || ETy == Type::UByteTy);
+  bool isString = (ETy == Type::Int8Ty || ETy == Type::Int8Ty);
 
   // Make sure the last character is a null char, as automatically added by C
   if (isString && (CPA->getNumOperands() == 0 ||
@@ -810,50 +806,19 @@
   case Type::BoolTyID:
 Out << (cast(CPV)->getValue() ? '1' : '0');
 break;
-  case Type::SByteTyID:
-  case Type::UByteTyID:
+  case Type::Int8TyID:
 Out << "((char)" << cast(CPV)->getSExtValue() << ")";
 break;
-  case Type::ShortTyID:
-  case Type::UShortTyID:
+  case Type::Int16TyID:
 Out << "((short)" << cast(CPV)->getSExtValue() << ")";
 break;
-  case Type::IntTyID:
-  case Type::UIntTyID:
+  case Type::Int32TyID:
 Out << "((int)" << cast(CPV)->getSExtValue() << ")";
 break;
-  case Type::LongTyID:
-  case Type::ULongTyID:
+  case Type::Int64TyID:
 Out << "((long long)" << cast(CPV)->getSExtValue() << "ll)";
 break;
 
-#if 0
-  case Type::IntTyID:
-if ((int)cast(CPV)->getSExtValue() == (int)0x8000)
-  Out << "((int)0x8000U)";   // Handle MININT specially to avoid 
warning
-else
-  Out << cast(CPV)->getSExtValue();
-break;
-
-  case Type::LongTyID:
-if (cast(CPV)->isMinValue(true))
-  Out << "(/*INT64_MIN*/(-9223372036854775807LL)-1)";
-else
-  Out << cast(CPV)->getSExtValue() << "ll";
-break;
-
-  case Type::UByteTyID:
-  case Type::UShortTyID:
-Out << cast(CPV)->getZExtValue();
-break;
-  case Type::UIntTyID:
-Out << cast(CPV)->getZExtValue() << 'u';
-break;
-  case Type::ULongTyID:
-Out << cast(CPV)->getZExtValue() << "ull";
-break;
-#endif
-
   case Type::FloatTyID:
   case Type::DoubleTyID: {
 ConstantFP *FPC = cast(CPV);
@@ -1627,10 +1592,8 @@
 void CWriter::printModuleTypes(const SymbolTable &ST) {
   Out << "/* Helper union for bitcasts */\n";
   Out << "typedef union {\n";
-  Out << "  unsigned int UInt;\n";
-  Out << "  signed int SInt;\n";
-  Out << "  unsigned long long ULong;\n";
-  Out << "  signed long long SLong;\n";
+  Out << "  unsigned int Int32;\n";
+  Out << "  unsigned long long Int64;\n";
   Out << "  float Float;\n";
   Out << "  double Double;\n";
   Out << "} llvmBitCastUnion;\n";
@@ -2060,8 +2023,7 @@
 
   // We must cast the results of binary operations which might be promoted.
   bool needsCast = false;
-  if ((I.getType() == Type::UByteTy) || (I.getType() == Type::SByteTy)
-  || (I.getType() == Type::UShortTy

[llvm-commits] CVS: llvm/lib/Target/TargetData.cpp

[Reid Spencer]

Changes in directory llvm/lib/Target:

TargetData.cpp updated: 1.73 -> 1.74
---
Log message:

For PR950: http://llvm.org/PR950 :
Three changes:
1. Convert signed integer types to signless versions.
2. Implement the @sext and @zext parameter attributes. Previously the
   type of an function parameter was used to determine whether it should
   be sign extended or zero extended before the call. This information is
   now communicated via the function type's parameter attributes. 
3. The interface to LowerCallTo had to be changed in order to accommodate
   the parameter attribute information. Although it would have been
   convenient to pass in the FunctionType itself, there isn't always one
   present in the caller. Consequently, a signedness indication for the
   result type and for each parameter was provided for in the interface
   to this method. All implementations were changed to make the adjustment
   necessary.


---
Diffs of the changes:  (+8 -12)

 TargetData.cpp |   20 
 1 files changed, 8 insertions(+), 12 deletions(-)


Index: llvm/lib/Target/TargetData.cpp
diff -u llvm/lib/Target/TargetData.cpp:1.73 llvm/lib/Target/TargetData.cpp:1.74
--- llvm/lib/Target/TargetData.cpp:1.73 Fri Dec  8 12:06:15 2006
+++ llvm/lib/Target/TargetData.cpp  Sat Dec 30 23:55:36 2006
@@ -243,14 +243,10 @@
   switch (Ty->getTypeID()) {
   case Type::BoolTyID:   Size = 1; Alignment = TD->getBoolAlignment(); return;
   case Type::VoidTyID:
-  case Type::UByteTyID:
-  case Type::SByteTyID:  Size = 1; Alignment = TD->getByteAlignment(); return;
-  case Type::UShortTyID:
-  case Type::ShortTyID:  Size = 2; Alignment = TD->getShortAlignment(); return;
-  case Type::UIntTyID:
-  case Type::IntTyID:Size = 4; Alignment = TD->getIntAlignment(); return;
-  case Type::ULongTyID:
-  case Type::LongTyID:   Size = 8; Alignment = TD->getLongAlignment(); return;
+  case Type::Int8TyID:   Size = 1; Alignment = TD->getByteAlignment(); return;
+  case Type::Int16TyID:  Size = 2; Alignment = TD->getShortAlignment(); return;
+  case Type::Int32TyID:  Size = 4; Alignment = TD->getIntAlignment(); return;
+  case Type::Int64TyID:  Size = 8; Alignment = TD->getLongAlignment(); return;
   case Type::FloatTyID:  Size = 4; Alignment = TD->getFloatAlignment(); return;
   case Type::DoubleTyID: Size = 8; Alignment = TD->getDoubleAlignment(); 
return;
   case Type::LabelTyID:
@@ -312,9 +308,9 @@
 const Type *TargetData::getIntPtrType() const {
   switch (getPointerSize()) {
   default: assert(0 && "Unknown pointer size!");
-  case 2: return Type::UShortTy;
-  case 4: return Type::UIntTy;
-  case 8: return Type::ULongTy;
+  case 2: return Type::Int16Ty;
+  case 4: return Type::Int32Ty;
+  case 8: return Type::Int64Ty;
   }
 }
 
@@ -329,7 +325,7 @@
 TI = gep_type_begin(ptrTy, Idx.begin(), Idx.end());
   for (unsigned CurIDX = 0; CurIDX != Idx.size(); ++CurIDX, ++TI) {
 if (const StructType *STy = dyn_cast(*TI)) {
-  assert(Idx[CurIDX]->getType() == Type::UIntTy && "Illegal struct idx");
+  assert(Idx[CurIDX]->getType() == Type::Int32Ty && "Illegal struct idx");
   unsigned FieldNo = cast(Idx[CurIDX])->getZExtValue();
 
   // Get structure layout information...



___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

[llvm-commits] CVS: llvm/lib/Target/IA64/IA64ISelLowering.cpp IA64ISelLowering.h

[Reid Spencer]

Changes in directory llvm/lib/Target/IA64:

IA64ISelLowering.cpp updated: 1.50 -> 1.51
IA64ISelLowering.h updated: 1.7 -> 1.8
---
Log message:

For PR950: http://llvm.org/PR950 :
Three changes:
1. Convert signed integer types to signless versions.
2. Implement the @sext and @zext parameter attributes. Previously the
   type of an function parameter was used to determine whether it should
   be sign extended or zero extended before the call. This information is
   now communicated via the function type's parameter attributes. 
3. The interface to LowerCallTo had to be changed in order to accommodate
   the parameter attribute information. Although it would have been
   convenient to pass in the FunctionType itself, there isn't always one
   present in the caller. Consequently, a signedness indication for the
   result type and for each parameter was provided for in the interface
   to this method. All implementations were changed to make the adjustment
   necessary.


---
Diffs of the changes:  (+20 -19)

 IA64ISelLowering.cpp |   32 +---
 IA64ISelLowering.h   |7 +++
 2 files changed, 20 insertions(+), 19 deletions(-)


Index: llvm/lib/Target/IA64/IA64ISelLowering.cpp
diff -u llvm/lib/Target/IA64/IA64ISelLowering.cpp:1.50 
llvm/lib/Target/IA64/IA64ISelLowering.cpp:1.51
--- llvm/lib/Target/IA64/IA64ISelLowering.cpp:1.50  Mon Nov 27 17:37:22 2006
+++ llvm/lib/Target/IA64/IA64ISelLowering.cpp   Sat Dec 30 23:55:36 2006
@@ -290,10 +290,10 @@
 
 std::pair
 IA64TargetLowering::LowerCallTo(SDOperand Chain,
-const Type *RetTy, bool isVarArg,
-unsigned CallingConv, bool isTailCall,
-SDOperand Callee, ArgListTy &Args,
-SelectionDAG &DAG) {
+const Type *RetTy, bool RetTyIsSigned, 
+bool isVarArg, unsigned CallingConv, 
+bool isTailCall, SDOperand Callee, 
+ArgListTy &Args, SelectionDAG &DAG) {
 
   MachineFunction &MF = DAG.getMachineFunction();
 
@@ -315,7 +315,8 @@
 std::max(outRegsUsed, MF.getInfo()->outRegsUsed);
 
   // keep stack frame 16-byte aligned
-  //assert(NumBytes==((NumBytes+15) & ~15) && "stack frame not 16-byte 
aligned!");
+  // assert(NumBytes==((NumBytes+15) & ~15) && 
+  //"stack frame not 16-byte aligned!");
   NumBytes = (NumBytes+15) & ~15;
   
   Chain = DAG.getCALLSEQ_START(Chain,DAG.getConstant(NumBytes, 
getPointerTy()));
@@ -328,7 +329,7 @@
   
   for (unsigned i = 0, e = Args.size(); i != e; ++i)
 {
-  SDOperand Val = Args[i].first;
+  SDOperand Val = Args[i].Node;
   MVT::ValueType ObjectVT = Val.getValueType();
   SDOperand ValToStore(0, 0), ValToConvert(0, 0);
   unsigned ObjSize=8;
@@ -337,14 +338,15 @@
   case MVT::i1:
   case MVT::i8:
   case MVT::i16:
-  case MVT::i32:
+  case MVT::i32: {
 //promote to 64-bits, sign/zero extending based on type
 //of the argument
-if(Args[i].second->isSigned())
-  Val = DAG.getNode(ISD::SIGN_EXTEND, MVT::i64, Val);
-else
-  Val = DAG.getNode(ISD::ZERO_EXTEND, MVT::i64, Val);
+ISD::NodeType ExtendKind = ISD::ZERO_EXTEND;
+if (Args[i].isSigned)
+  ExtendKind = ISD::SIGN_EXTEND;
+Val = DAG.getNode(ExtendKind, MVT::i64, Val);
 // XXX: fall through
+  }
   case MVT::i64:
 //ObjSize = 8;
 if(RegValuesToPass.size() >= 8) {
@@ -422,7 +424,8 @@
   unsigned seenConverts = 0;
   for (unsigned i = 0, e = RegValuesToPass.size(); i != e; ++i) {
 if(MVT::isFloatingPoint(RegValuesToPass[i].getValueType())) {
-  Chain = DAG.getCopyToReg(Chain, IntArgRegs[i], Converts[seenConverts++], 
InFlag);
+  Chain = DAG.getCopyToReg(Chain, IntArgRegs[i], Converts[seenConverts++], 
+   InFlag);
   InFlag = Chain.getValue(1);
 }
   }
@@ -432,8 +435,7 @@
   for (unsigned i = 0, e = RegValuesToPass.size(); i != e; ++i) {
 Chain = DAG.getCopyToReg(Chain,
   MVT::isInteger(RegValuesToPass[i].getValueType()) ?
-  IntArgRegs[i] : 
FPArgRegs[usedFPArgs++],
-  RegValuesToPass[i], InFlag);
+IntArgRegs[i] : FPArgRegs[usedFPArgs++], RegValuesToPass[i], InFlag);
 InFlag = Chain.getValue(1);
   }
 
@@ -483,7 +485,7 @@
 case MVT::i1: { // bools are just like other integers (returned in r8)
   // we *could* fall through to the truncate below, but this saves a
   // few redundant predicate ops
-  SDOperand boolInR8 = DAG.getCopyFromReg(Chain, IA64::r8, MVT::i64, 
InFlag);
+  SDOperand boolInR8 = DAG.getCopyFromReg(Chain, IA64::r8, 
MVT::i64,InFlag);
   InFlag = boolInR8.getValue(2);
   Chain = boolInR8.getValue(1);
   SDOperand zeroReg = DAG.getCopyFromReg(Chain, IA64::r0, MVT::i64, 
InFlag);


I

[llvm-commits] CVS: llvm/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp SelectionDAGISel.cpp

[Reid Spencer]

Changes in directory llvm/lib/CodeGen/SelectionDAG:

LegalizeDAG.cpp updated: 1.449 -> 1.450
SelectionDAGISel.cpp updated: 1.334 -> 1.335
---
Log message:

For PR950: http://llvm.org/PR950 :
Three changes:
1. Convert signed integer types to signless versions.
2. Implement the @sext and @zext parameter attributes. Previously the
   type of an function parameter was used to determine whether it should
   be sign extended or zero extended before the call. This information is
   now communicated via the function type's parameter attributes. 
3. The interface to LowerCallTo had to be changed in order to accommodate
   the parameter attribute information. Although it would have been
   convenient to pass in the FunctionType itself, there isn't always one
   present in the caller. Consequently, a signedness indication for the
   result type and for each parameter was provided for in the interface
   to this method. All implementations were changed to make the adjustment
   necessary.


---
Diffs of the changes:  (+119 -97)

 LegalizeDAG.cpp  |  108 ++-
 SelectionDAGISel.cpp |  108 +--
 2 files changed, 119 insertions(+), 97 deletions(-)


Index: llvm/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp
diff -u llvm/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp:1.449 
llvm/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp:1.450
--- llvm/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp:1.449 Mon Dec 18 19:44:04 2006
+++ llvm/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp   Sat Dec 30 23:55:36 2006
@@ -186,7 +186,7 @@
 
   SDOperand CreateStackTemporary(MVT::ValueType VT);
 
-  SDOperand ExpandLibCall(const char *Name, SDNode *Node,
+  SDOperand ExpandLibCall(const char *Name, SDNode *Node, bool isSigned,
   SDOperand &Hi);
   SDOperand ExpandIntToFP(bool isSigned, MVT::ValueType DestTy,
   SDOperand Source);
@@ -2122,33 +2122,42 @@
   // operation to an explicit libcall as appropriate.
   MVT::ValueType IntPtr = TLI.getPointerTy();
   const Type *IntPtrTy = TLI.getTargetData()->getIntPtrType();
-  std::vector > Args;
+  TargetLowering::ArgListTy Args;
+  TargetLowering::ArgListEntry Entry;
 
   const char *FnName = 0;
   if (Node->getOpcode() == ISD::MEMSET) {
-Args.push_back(std::make_pair(Tmp2, IntPtrTy));
+Entry.Node = Tmp2;
+Entry.Ty = IntPtrTy;
+Entry.isSigned = false;
+Args.push_back(Entry);
 // Extend the (previously legalized) ubyte argument to be an int value
 // for the call.
 if (Tmp3.getValueType() > MVT::i32)
   Tmp3 = DAG.getNode(ISD::TRUNCATE, MVT::i32, Tmp3);
 else
   Tmp3 = DAG.getNode(ISD::ZERO_EXTEND, MVT::i32, Tmp3);
-Args.push_back(std::make_pair(Tmp3, Type::IntTy));
-Args.push_back(std::make_pair(Tmp4, IntPtrTy));
+Entry.Node = Tmp3; Entry.Ty = Type::Int32Ty; Entry.isSigned = true;
+Args.push_back(Entry);
+Entry.Node = Tmp4; Entry.Ty = IntPtrTy; Entry.isSigned = false;
+Args.push_back(Entry);
 
 FnName = "memset";
   } else if (Node->getOpcode() == ISD::MEMCPY ||
  Node->getOpcode() == ISD::MEMMOVE) {
-Args.push_back(std::make_pair(Tmp2, IntPtrTy));
-Args.push_back(std::make_pair(Tmp3, IntPtrTy));
-Args.push_back(std::make_pair(Tmp4, IntPtrTy));
+Entry.Node = Tmp2; Entry.Ty = IntPtrTy; Entry.isSigned = false;
+Args.push_back(Entry);
+Entry.Node = Tmp3; Entry.Ty = IntPtrTy; Entry.isSigned = false;
+Args.push_back(Entry);
+Entry.Node = Tmp4; Entry.Ty = IntPtrTy; Entry.isSigned = false;
+Args.push_back(Entry);
 FnName = Node->getOpcode() == ISD::MEMMOVE ? "memmove" : "memcpy";
   } else {
 assert(0 && "Unknown op!");
   }
 
   std::pair CallResult =
-TLI.LowerCallTo(Tmp1, Type::VoidTy, false, CallingConv::C, false,
+TLI.LowerCallTo(Tmp1, Type::VoidTy, false, false, CallingConv::C, 
false,
 DAG.getExternalSymbol(FnName, IntPtr), Args, DAG);
   Result = CallResult.second;
   break;
@@ -2243,7 +2252,8 @@
   const char *FnName = Node->getOpcode() == ISD::UDIV
 ? "__udivsi3" : "__divsi3";
   SDOperand Dummy;
-  Result = ExpandLibCall(FnName, Node, Dummy);
+  bool isSigned = Node->getOpcode() == ISD::SDIV;
+  Result = ExpandLibCall(FnName, Node, isSigned, Dummy);
 };
 break;
   }
@@ -2346,7 +2356,7 @@
DAG.getNode(ISD::FP_EXTEND, MVT::f64, 
Tmp2));
   }
   SDOperand Dummy;
-  Result = ExpandLibCall(FnName, Node, Dummy);
+  Result = ExpandLibCall(FnName, Node, false, Dummy);
   break;
 }
 break;
@@ -2419,6 +2429,7 @@
   break;
 case TargetLowering::Expand:
   unsigned DivOpc= (Node->getOpcode() == ISD::URE

[llvm-commits] CVS: llvm/lib/Target/X86/X86ISelLowering.cpp X86TargetAsmInfo.cpp

[Reid Spencer]

Changes in directory llvm/lib/Target/X86:

X86ISelLowering.cpp updated: 1.307 -> 1.308
X86TargetAsmInfo.cpp updated: 1.13 -> 1.14
---
Log message:

For PR950: http://llvm.org/PR950 :
Three changes:
1. Convert signed integer types to signless versions.
2. Implement the @sext and @zext parameter attributes. Previously the
   type of an function parameter was used to determine whether it should
   be sign extended or zero extended before the call. This information is
   now communicated via the function type's parameter attributes. 
3. The interface to LowerCallTo had to be changed in order to accommodate
   the parameter attribute information. Although it would have been
   convenient to pass in the FunctionType itself, there isn't always one
   present in the caller. Consequently, a signedness indication for the
   result type and for each parameter was provided for in the interface
   to this method. All implementations were changed to make the adjustment
   necessary.


---
Diffs of the changes:  (+25 -17)

 X86ISelLowering.cpp  |   32 
 X86TargetAsmInfo.cpp |   10 +-
 2 files changed, 25 insertions(+), 17 deletions(-)


Index: llvm/lib/Target/X86/X86ISelLowering.cpp
diff -u llvm/lib/Target/X86/X86ISelLowering.cpp:1.307 
llvm/lib/Target/X86/X86ISelLowering.cpp:1.308
--- llvm/lib/Target/X86/X86ISelLowering.cpp:1.307   Fri Dec 22 16:29:05 2006
+++ llvm/lib/Target/X86/X86ISelLowering.cpp Sat Dec 30 23:55:36 2006
@@ -4447,14 +4447,21 @@
   (I && I->getValue() < Subtarget->getMinRepStrSizeThreshold())) {
 MVT::ValueType IntPtr = getPointerTy();
 const Type *IntPtrTy = getTargetData()->getIntPtrType();
-std::vector > Args;
-Args.push_back(std::make_pair(Op.getOperand(1), IntPtrTy));
+TargetLowering::ArgListTy Args; 
+TargetLowering::ArgListEntry Entry;
+Entry.Node = Op.getOperand(1);
+Entry.Ty = IntPtrTy;
+Entry.isSigned = false;
+Args.push_back(Entry);
 // Extend the ubyte argument to be an int value for the call.
-SDOperand Val = DAG.getNode(ISD::ZERO_EXTEND, MVT::i32, Op.getOperand(2));
-Args.push_back(std::make_pair(Val, IntPtrTy));
-Args.push_back(std::make_pair(Op.getOperand(3), IntPtrTy));
+Entry.Node = DAG.getNode(ISD::ZERO_EXTEND, MVT::i32, Op.getOperand(2));
+Entry.Ty = IntPtrTy;
+Entry.isSigned = false;
+Args.push_back(Entry);
+Entry.Node = Op.getOperand(3);
+Args.push_back(Entry);
 std::pair CallResult =
-  LowerCallTo(Chain, Type::VoidTy, false, CallingConv::C, false,
+  LowerCallTo(Chain, Type::VoidTy, false, false, CallingConv::C, false,
   DAG.getExternalSymbol("memset", IntPtr), Args, DAG);
 return CallResult.second;
   }
@@ -4601,13 +4608,14 @@
   if ((Align & 3) != 0 ||
   (I && I->getValue() < Subtarget->getMinRepStrSizeThreshold())) {
 MVT::ValueType IntPtr = getPointerTy();
-const Type *IntPtrTy = getTargetData()->getIntPtrType();
-std::vector > Args;
-Args.push_back(std::make_pair(Op.getOperand(1), IntPtrTy));
-Args.push_back(std::make_pair(Op.getOperand(2), IntPtrTy));
-Args.push_back(std::make_pair(Op.getOperand(3), IntPtrTy));
+TargetLowering::ArgListTy Args;
+TargetLowering::ArgListEntry Entry;
+Entry.Ty = getTargetData()->getIntPtrType(); Entry.isSigned = false;
+Entry.Node = Op.getOperand(1); Args.push_back(Entry);
+Entry.Node = Op.getOperand(2); Args.push_back(Entry);
+Entry.Node = Op.getOperand(3); Args.push_back(Entry);
 std::pair CallResult =
-  LowerCallTo(Chain, Type::VoidTy, false, CallingConv::C, false,
+  LowerCallTo(Chain, Type::VoidTy, false, false, CallingConv::C, false,
   DAG.getExternalSymbol("memcpy", IntPtr), Args, DAG);
 return CallResult.second;
   }


Index: llvm/lib/Target/X86/X86TargetAsmInfo.cpp
diff -u llvm/lib/Target/X86/X86TargetAsmInfo.cpp:1.13 
llvm/lib/Target/X86/X86TargetAsmInfo.cpp:1.14
--- llvm/lib/Target/X86/X86TargetAsmInfo.cpp:1.13   Tue Dec 19 15:04:20 2006
+++ llvm/lib/Target/X86/X86TargetAsmInfo.cppSat Dec 30 23:55:36 2006
@@ -172,13 +172,13 @@
   !CI->getType()->isInteger())
 return false;
   
-  const Type *Ty = CI->getType()->getUnsignedVersion();
+  const Type *Ty = CI->getType();
   const char *IntName;
   switch (Ty->getTypeID()) {
   default: return false;
-  case Type::UShortTyID: IntName = "llvm.bswap.i16"; break;
-  case Type::UIntTyID:   IntName = "llvm.bswap.i32"; break;
-  case Type::ULongTyID:  IntName = "llvm.bswap.i64"; break;
+  case Type::Int16TyID: IntName = "llvm.bswap.i16"; break;
+  case Type::Int32TyID:   IntName = "llvm.bswap.i32"; break;
+  case Type::Int64TyID:  IntName = "llvm.bswap.i64"; break;
   }
 
   // Okay, we can do this xform, do so now.
@@ -226,7 +226,7 @@
 }
 break;
   case 3:
-if (CI->getType() == Type::ULongTy && Constraints.size() >= 2 &&
+if (CI->getType() == Type::Int64Ty && Constraints.size() >= 2 &&
 Constraints[0].C

[llvm-commits] CVS: llvm/lib/Target/Alpha/AlphaISelDAGToDAG.cpp AlphaISelLowering.cpp AlphaISelLowering.h

[Reid Spencer]

Changes in directory llvm/lib/Target/Alpha:

AlphaISelDAGToDAG.cpp updated: 1.64 -> 1.65
AlphaISelLowering.cpp updated: 1.75 -> 1.76
AlphaISelLowering.h updated: 1.22 -> 1.23
---
Log message:

For PR950: http://llvm.org/PR950 :
Three changes:
1. Convert signed integer types to signless versions.
2. Implement the @sext and @zext parameter attributes. Previously the
   type of an function parameter was used to determine whether it should
   be sign extended or zero extended before the call. This information is
   now communicated via the function type's parameter attributes. 
3. The interface to LowerCallTo had to be changed in order to accommodate
   the parameter attribute information. Although it would have been
   convenient to pass in the FunctionType itself, there isn't always one
   present in the caller. Consequently, a signedness indication for the
   result type and for each parameter was provided for in the interface
   to this method. All implementations were changed to make the adjustment
   necessary.


---
Diffs of the changes:  (+12 -12)

 AlphaISelDAGToDAG.cpp |2 +-
 AlphaISelLowering.cpp |   16 
 AlphaISelLowering.h   |6 +++---
 3 files changed, 12 insertions(+), 12 deletions(-)


Index: llvm/lib/Target/Alpha/AlphaISelDAGToDAG.cpp
diff -u llvm/lib/Target/Alpha/AlphaISelDAGToDAG.cpp:1.64 
llvm/lib/Target/Alpha/AlphaISelDAGToDAG.cpp:1.65
--- llvm/lib/Target/Alpha/AlphaISelDAGToDAG.cpp:1.64Tue Dec 19 16:59:25 2006
+++ llvm/lib/Target/Alpha/AlphaISelDAGToDAG.cpp Sat Dec 30 23:55:36 2006
@@ -322,7 +322,7 @@
   //   val32 >= IMM_LOW  + IMM_LOW  * IMM_MULT) //always true
   break; //(zext (LDAH (LDA)))
 //Else use the constant pool
-ConstantInt *C = ConstantInt::get(Type::ULongTy, uval);
+ConstantInt *C = ConstantInt::get(Type::Int64Ty, uval);
 SDOperand CPI = CurDAG->getTargetConstantPool(C, MVT::i64);
 SDNode *Tmp = CurDAG->getTargetNode(Alpha::LDAHr, MVT::i64, CPI,
 getGlobalBaseReg());


Index: llvm/lib/Target/Alpha/AlphaISelLowering.cpp
diff -u llvm/lib/Target/Alpha/AlphaISelLowering.cpp:1.75 
llvm/lib/Target/Alpha/AlphaISelLowering.cpp:1.76
--- llvm/lib/Target/Alpha/AlphaISelLowering.cpp:1.75Thu Dec  7 16:21:48 2006
+++ llvm/lib/Target/Alpha/AlphaISelLowering.cpp Sat Dec 30 23:55:36 2006
@@ -317,8 +317,8 @@
 }
 
 std::pair
-AlphaTargetLowering::LowerCallTo(SDOperand Chain,
- const Type *RetTy, bool isVarArg,
+AlphaTargetLowering::LowerCallTo(SDOperand Chain, const Type *RetTy, 
+ bool RetTyIsSigned, bool isVarArg,
  unsigned CallingConv, bool isTailCall,
  SDOperand Callee, ArgListTy &Args,
  SelectionDAG &DAG) {
@@ -331,7 +331,7 @@
   std::vector args_to_use;
   for (unsigned i = 0, e = Args.size(); i != e; ++i)
   {
-switch (getValueType(Args[i].second)) {
+switch (getValueType(Args[i].Ty)) {
 default: assert(0 && "Unexpected ValueType for argument!");
 case MVT::i1:
 case MVT::i8:
@@ -339,17 +339,17 @@
 case MVT::i32:
   // Promote the integer to 64 bits.  If the input type is signed use a
   // sign extend, otherwise use a zero extend.
-  if (Args[i].second->isSigned())
-Args[i].first = DAG.getNode(ISD::SIGN_EXTEND, MVT::i64, Args[i].first);
+  if (Args[i].isSigned)
+Args[i].Node = DAG.getNode(ISD::SIGN_EXTEND, MVT::i64, Args[i].Node);
   else
-Args[i].first = DAG.getNode(ISD::ZERO_EXTEND, MVT::i64, Args[i].first);
+Args[i].Node = DAG.getNode(ISD::ZERO_EXTEND, MVT::i64, Args[i].Node);
   break;
 case MVT::i64:
 case MVT::f64:
 case MVT::f32:
   break;
 }
-args_to_use.push_back(Args[i].first);
+args_to_use.push_back(Args[i].Node);
   }
 
   std::vector RetVals;
@@ -373,7 +373,7 @@
   SDOperand RetVal = TheCall;
 
   if (RetTyVT != ActualRetTyVT) {
-RetVal = DAG.getNode(RetTy->isSigned() ? ISD::AssertSext : ISD::AssertZext,
+RetVal = DAG.getNode(RetTyIsSigned ? ISD::AssertSext : ISD::AssertZext,
  MVT::i64, RetVal, DAG.getValueType(RetTyVT));
 RetVal = DAG.getNode(ISD::TRUNCATE, RetTyVT, RetVal);
   }


Index: llvm/lib/Target/Alpha/AlphaISelLowering.h
diff -u llvm/lib/Target/Alpha/AlphaISelLowering.h:1.22 
llvm/lib/Target/Alpha/AlphaISelLowering.h:1.23
--- llvm/lib/Target/Alpha/AlphaISelLowering.h:1.22  Tue Oct 31 10:49:55 2006
+++ llvm/lib/Target/Alpha/AlphaISelLowering.h   Sat Dec 30 23:55:36 2006
@@ -77,9 +77,9 @@
 /// LowerCallTo - This hook lowers an abstract call to a function into an
 /// actual call.
 virtual std::pair
-LowerCallTo(SDOperand Chain, const Type *RetTy, bool isVarArg, unsigned CC,
-bool isTailCall, SDOperand Callee, ArgListTy &Args,
-SelectionDAG &DAG);
+LowerCallTo(SDOperand Chain, const Type *RetTy, bool RetT

[llvm-commits] CVS: llvm/test/Regression/Transforms/Inline/casts.ll

[Reid Spencer]

Changes in directory llvm/test/Regression/Transforms/Inline:

casts.ll updated: 1.2 -> 1.3
---
Log message:

For PR950: http://llvm.org/PR950 :
Update the test suite to accommodate the change from signed integer types
to signless integer types. The changes were of only a few kinds:

1. Make sure llvm-upgrade is run on the source which does the bulk of the
   changes automatically.

2. Change things like "grep 'int'" to "grep 'i32'"

3. In several tests bitcasting caused the same name to be reused in the
   same type plane. These had to be manually fixed. The fix was (generally)
   to leave the bitcast and provide the instruction with a new name. This
   should not affect the semantics of the test. In a few cases, the 
   bitcasts were known to be superfluous and irrelevant to the test case
   so they were removed.

4. One test case uses a bytecode file which needed to be updated to the
   latest bytecode format.



---
Diffs of the changes:  (+1 -1)

 casts.ll |2 +-
 1 files changed, 1 insertion(+), 1 deletion(-)


Index: llvm/test/Regression/Transforms/Inline/casts.ll
diff -u llvm/test/Regression/Transforms/Inline/casts.ll:1.2 
llvm/test/Regression/Transforms/Inline/casts.ll:1.3
--- llvm/test/Regression/Transforms/Inline/casts.ll:1.2 Fri Dec  1 22:23:09 2006
+++ llvm/test/Regression/Transforms/Inline/casts.ll Sun Dec 31 00:01:59 2006
@@ -1,4 +1,4 @@
-; RUN: llvm-upgrade < %s | llvm-as | opt -inline | llvm-dis | grep 'ret int 1'
+; RUN: llvm-upgrade < %s | llvm-as | opt -inline | llvm-dis | grep 'ret i32 1'
 ; ModuleID = 'short.opt.bc'
 
 implementation   ; Functions:



___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

[llvm-commits] CVS: llvm/test/Regression/Transforms/ArgumentPromotion/control-flow2.ll

[Reid Spencer]

Changes in directory llvm/test/Regression/Transforms/ArgumentPromotion:

control-flow2.ll updated: 1.2 -> 1.3
---
Log message:

For PR950: http://llvm.org/PR950 :
Update the test suite to accommodate the change from signed integer types
to signless integer types. The changes were of only a few kinds:

1. Make sure llvm-upgrade is run on the source which does the bulk of the
   changes automatically.

2. Change things like "grep 'int'" to "grep 'i32'"

3. In several tests bitcasting caused the same name to be reused in the
   same type plane. These had to be manually fixed. The fix was (generally)
   to leave the bitcast and provide the instruction with a new name. This
   should not affect the semantics of the test. In a few cases, the 
   bitcasts were known to be superfluous and irrelevant to the test case
   so they were removed.

4. One test case uses a bytecode file which needed to be updated to the
   latest bytecode format.



---
Diffs of the changes:  (+1 -1)

 control-flow2.ll |2 +-
 1 files changed, 1 insertion(+), 1 deletion(-)


Index: llvm/test/Regression/Transforms/ArgumentPromotion/control-flow2.ll
diff -u llvm/test/Regression/Transforms/ArgumentPromotion/control-flow2.ll:1.2 
llvm/test/Regression/Transforms/ArgumentPromotion/control-flow2.ll:1.3
--- llvm/test/Regression/Transforms/ArgumentPromotion/control-flow2.ll:1.2  
Fri Dec  1 22:23:08 2006
+++ llvm/test/Regression/Transforms/ArgumentPromotion/control-flow2.ll  Sun Dec 
31 00:01:59 2006
@@ -1,4 +1,4 @@
-; RUN: llvm-upgrade < %s | llvm-as | opt -argpromotion | llvm-dis | grep 'load 
int\* %A'
+; RUN: llvm-upgrade < %s | llvm-as | opt -argpromotion | llvm-dis | grep 'load 
i32\* %A'
 
 implementation
 



___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

[llvm-commits] CVS: llvm/test/Regression/CodeGen/X86/2006-05-02-InstrSched1.ll 2006-08-07-CycleInDAG.ll loop-hoist.ll loop-strength-reduce.ll trunc-to-bool.ll vec_ins_extract.ll

[Reid Spencer]

Changes in directory llvm/test/Regression/CodeGen/X86:

2006-05-02-InstrSched1.ll updated: 1.2 -> 1.3
2006-08-07-CycleInDAG.ll updated: 1.2 -> 1.3
loop-hoist.ll updated: 1.3 -> 1.4
loop-strength-reduce.ll updated: 1.3 -> 1.4
trunc-to-bool.ll updated: 1.4 -> 1.5
vec_ins_extract.ll updated: 1.3 -> 1.4
---
Log message:

For PR950: http://llvm.org/PR950 :
Update the test suite to accommodate the change from signed integer types
to signless integer types. The changes were of only a few kinds:

1. Make sure llvm-upgrade is run on the source which does the bulk of the
   changes automatically.

2. Change things like "grep 'int'" to "grep 'i32'"

3. In several tests bitcasting caused the same name to be reused in the
   same type plane. These had to be manually fixed. The fix was (generally)
   to leave the bitcast and provide the instruction with a new name. This
   should not affect the semantics of the test. In a few cases, the 
   bitcasts were known to be superfluous and irrelevant to the test case
   so they were removed.

4. One test case uses a bytecode file which needed to be updated to the
   latest bytecode format.



---
Diffs of the changes:  (+38 -37)

 2006-05-02-InstrSched1.ll |6 +
 2006-08-07-CycleInDAG.ll  |4 +--
 loop-hoist.ll |4 +--
 loop-strength-reduce.ll   |6 ++---
 trunc-to-bool.ll  |   51 --
 vec_ins_extract.ll|4 +--
 6 files changed, 38 insertions(+), 37 deletions(-)


Index: llvm/test/Regression/CodeGen/X86/2006-05-02-InstrSched1.ll
diff -u llvm/test/Regression/CodeGen/X86/2006-05-02-InstrSched1.ll:1.2 
llvm/test/Regression/CodeGen/X86/2006-05-02-InstrSched1.ll:1.3
--- llvm/test/Regression/CodeGen/X86/2006-05-02-InstrSched1.ll:1.2  Fri Dec 
 1 22:23:08 2006
+++ llvm/test/Regression/CodeGen/X86/2006-05-02-InstrSched1.ll  Sun Dec 31 
00:01:59 2006
@@ -11,10 +11,8 @@
%tmp4 = getelementptr ubyte* %tmp, uint %tmp3   ;  
[#uses=1]
%tmp7 = load uint* %tmp ;  [#uses=1]
%tmp8 = getelementptr ubyte* %tmp, uint %tmp7   ;  
[#uses=1]
-   %tmp8 = cast ubyte* %tmp8 to sbyte* ;  [#uses=1]
-   %tmp4 = cast ubyte* %tmp4 to sbyte* ;  [#uses=1]
-   %tmp = tail call int %memcmp( sbyte* %tmp8, sbyte* %tmp4, uint %tmp )   
;  [#uses=1]
-   ret int %tmp
+   %result = tail call int %memcmp( sbyte* %tmp8, sbyte* %tmp4, uint %tmp 
);  [#uses=1]
+   ret int %result
 }
 
 declare int %memcmp(sbyte*, sbyte*, uint)


Index: llvm/test/Regression/CodeGen/X86/2006-08-07-CycleInDAG.ll
diff -u llvm/test/Regression/CodeGen/X86/2006-08-07-CycleInDAG.ll:1.2 
llvm/test/Regression/CodeGen/X86/2006-08-07-CycleInDAG.ll:1.3
--- llvm/test/Regression/CodeGen/X86/2006-08-07-CycleInDAG.ll:1.2   Fri Dec 
 1 22:23:08 2006
+++ llvm/test/Regression/CodeGen/X86/2006-08-07-CycleInDAG.ll   Sun Dec 31 
00:01:59 2006
@@ -10,8 +10,8 @@
%tmp24.i = load int* null   ;  [#uses=1]
%tmp13.i12.i = tail call double %ldexp( double 0.00e+00, int 0 )
;  [#uses=1]
%tmp13.i13.i = cast double %tmp13.i12.i to float; 
 [#uses=1]
-   %tmp11.i = load int* null   ;  [#uses=1]
-   %tmp11.i = cast int %tmp11.i to uint;  [#uses=1]
+   %tmp11.s = load int* null   ;  [#uses=1]
+   %tmp11.i = cast int %tmp11.s to uint;  [#uses=1]
%n.i = cast int %tmp24.i to uint;  [#uses=1]
%tmp13.i7 = mul uint %tmp11.i, %n.i ;  [#uses=1]
%tmp.i8 = tail call sbyte* %calloc( uint %tmp13.i7, uint 4 )
;  [#uses=0]


Index: llvm/test/Regression/CodeGen/X86/loop-hoist.ll
diff -u llvm/test/Regression/CodeGen/X86/loop-hoist.ll:1.3 
llvm/test/Regression/CodeGen/X86/loop-hoist.ll:1.4
--- llvm/test/Regression/CodeGen/X86/loop-hoist.ll:1.3  Sat Dec  2 14:39:25 2006
+++ llvm/test/Regression/CodeGen/X86/loop-hoist.ll  Sun Dec 31 00:01:59 2006
@@ -5,9 +5,9 @@
 
 implementation   ; Functions:
 
-void %foo(int %N) {
+void %foo(int %N.in) {
 entry:
-%N = cast int %N to uint;  [#uses=1]
+%N = cast int %N.in to uint;  [#uses=1]
 br label %cond_true
 
 cond_true:  ; preds = %cond_true, %entry


Index: llvm/test/Regression/CodeGen/X86/loop-strength-reduce.ll
diff -u llvm/test/Regression/CodeGen/X86/loop-strength-reduce.ll:1.3 
llvm/test/Regression/CodeGen/X86/loop-strength-reduce.ll:1.4
--- llvm/test/Regression/CodeGen/X86/loop-strength-reduce.ll:1.3Fri Dec 
 1 22:23:08 2006
+++ llvm/test/Regression/CodeGen/X86/loop-strength-reduce.llSun Dec 31 
00:01:59 2006
@@ -4,10 +4,10 @@
 
 %A = internal global [16 x [16 x int]] zeroinitializer, align 32
 
-void %test(int %row, int %N) {
+void %test(int %row, int %N.in) {
 entry:
-   %N = cast int %N to uint
-   %tmp5 = setgt int %N, 0
+   %N = cast int %N.i

[llvm-commits] CVS: llvm/tools/bugpoint/ExtractFunction.cpp Miscompilation.cpp

[Reid Spencer]

Changes in directory llvm/tools/bugpoint:

ExtractFunction.cpp updated: 1.55 -> 1.56
Miscompilation.cpp updated: 1.82 -> 1.83
---
Log message:

For PR950: http://llvm.org/PR950 :
Convert signed integer types to signless.


---
Diffs of the changes:  (+4 -4)

 ExtractFunction.cpp |2 +-
 Miscompilation.cpp  |6 +++---
 2 files changed, 4 insertions(+), 4 deletions(-)


Index: llvm/tools/bugpoint/ExtractFunction.cpp
diff -u llvm/tools/bugpoint/ExtractFunction.cpp:1.55 
llvm/tools/bugpoint/ExtractFunction.cpp:1.56
--- llvm/tools/bugpoint/ExtractFunction.cpp:1.55Sun Nov 26 19:05:10 2006
+++ llvm/tools/bugpoint/ExtractFunction.cpp Sun Dec 31 00:02:26 2006
@@ -180,7 +180,7 @@
   std::vector ArrayElts;
   for (unsigned i = 0, e = TorList.size(); i != e; ++i) {
 std::vector Elts;
-Elts.push_back(ConstantInt::get(Type::IntTy, TorList[i].second));
+Elts.push_back(ConstantInt::get(Type::Int32Ty, TorList[i].second));
 Elts.push_back(TorList[i].first);
 ArrayElts.push_back(ConstantStruct::get(Elts));
   }


Index: llvm/tools/bugpoint/Miscompilation.cpp
diff -u llvm/tools/bugpoint/Miscompilation.cpp:1.82 
llvm/tools/bugpoint/Miscompilation.cpp:1.83
--- llvm/tools/bugpoint/Miscompilation.cpp:1.82 Sat Dec 23 00:05:41 2006
+++ llvm/tools/bugpoint/Miscompilation.cpp  Sun Dec 31 00:02:26 2006
@@ -675,8 +675,8 @@
   // Prototype: void *getPointerToNamedFunction(const char* Name)
   Function *resolverFunc =
 Safe->getOrInsertFunction("getPointerToNamedFunction",
-  PointerType::get(Type::SByteTy),
-  PointerType::get(Type::SByteTy), (Type *)0);
+  PointerType::get(Type::Int8Ty),
+  PointerType::get(Type::Int8Ty), (Type *)0);
 
   // Use the function we just added to get addresses of functions we need.
   for (Module::iterator F = Safe->begin(), E = Safe->end(); F != E; ++F) {
@@ -697,7 +697,7 @@
 // sbyte* so it matches the signature of the resolver function.
 
 // GetElementPtr *funcName, ulong 0, ulong 0
-std::vector GEPargs(2,Constant::getNullValue(Type::IntTy));
+std::vector 
GEPargs(2,Constant::getNullValue(Type::Int32Ty));
 Value *GEP =
   ConstantExpr::getGetElementPtr(funcName, GEPargs);
 std::vector ResolverArgs;



___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

[llvm-commits] CVS: llvm/test/Regression/Transforms/Reassociate/2005-09-01-ArrayOutOfBounds.ll shifttest.ll

[Reid Spencer]

Changes in directory llvm/test/Regression/Transforms/Reassociate:

2005-09-01-ArrayOutOfBounds.ll updated: 1.2 -> 1.3
shifttest.ll updated: 1.2 -> 1.3
---
Log message:

For PR950: http://llvm.org/PR950 :
Update the test suite to accommodate the change from signed integer types
to signless integer types. The changes were of only a few kinds:

1. Make sure llvm-upgrade is run on the source which does the bulk of the
   changes automatically.

2. Change things like "grep 'int'" to "grep 'i32'"

3. In several tests bitcasting caused the same name to be reused in the
   same type plane. These had to be manually fixed. The fix was (generally)
   to leave the bitcast and provide the instruction with a new name. This
   should not affect the semantics of the test. In a few cases, the 
   bitcasts were known to be superfluous and irrelevant to the test case
   so they were removed.

4. One test case uses a bytecode file which needed to be updated to the
   latest bytecode format.



---
Diffs of the changes:  (+2 -2)

 2005-09-01-ArrayOutOfBounds.ll |2 +-
 shifttest.ll   |2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)


Index: 
llvm/test/Regression/Transforms/Reassociate/2005-09-01-ArrayOutOfBounds.ll
diff -u 
llvm/test/Regression/Transforms/Reassociate/2005-09-01-ArrayOutOfBounds.ll:1.2 
llvm/test/Regression/Transforms/Reassociate/2005-09-01-ArrayOutOfBounds.ll:1.3
--- 
llvm/test/Regression/Transforms/Reassociate/2005-09-01-ArrayOutOfBounds.ll:1.2  
Fri Dec  1 22:23:10 2006
+++ llvm/test/Regression/Transforms/Reassociate/2005-09-01-ArrayOutOfBounds.ll  
Sun Dec 31 00:01:59 2006
@@ -1,4 +1,4 @@
-; RUN: llvm-upgrade < %s | llvm-as | opt -reassociate -instcombine | llvm-dis 
| grep 'ret int 0'
+; RUN: llvm-upgrade < %s | llvm-as | opt -reassociate -instcombine | llvm-dis 
| grep 'ret i32 0'
 
 int %f(int %a0, int %a1, int %a2, int %a3, int %a4) {
%tmp.2 = add int %a4, %a3   ;  [#uses=1]


Index: llvm/test/Regression/Transforms/Reassociate/shifttest.ll
diff -u llvm/test/Regression/Transforms/Reassociate/shifttest.ll:1.2 
llvm/test/Regression/Transforms/Reassociate/shifttest.ll:1.3
--- llvm/test/Regression/Transforms/Reassociate/shifttest.ll:1.2Fri Dec 
 1 22:23:10 2006
+++ llvm/test/Regression/Transforms/Reassociate/shifttest.llSun Dec 31 
00:01:59 2006
@@ -1,6 +1,6 @@
 ; With shl->mul reassociation, we can see that this is (shl A, 9) * A
 ;
-; RUN: llvm-upgrade < %s | llvm-as | opt -reassociate -instcombine | llvm-dis 
| grep 'shl .*, ubyte 9'
+; RUN: llvm-upgrade < %s | llvm-as | opt -reassociate -instcombine | llvm-dis 
| grep 'shl .*, i8 9'
 
 int %test(int %A, int %B) {
%X = shl int %A, ubyte 5



___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

[llvm-commits] CVS: llvm/test/Regression/CFrontend/2005-12-04-DeclarationLineNumbers.c 2006-01-16-BitCountIntrinsicsUnsigned.c 2006-03-03-MissingInitializer.c

[Reid Spencer]

Changes in directory llvm/test/Regression/CFrontend:

2005-12-04-DeclarationLineNumbers.c updated: 1.4 -> 1.5
2006-01-16-BitCountIntrinsicsUnsigned.c updated: 1.1 -> 1.2
2006-03-03-MissingInitializer.c updated: 1.1 -> 1.2
---
Log message:

For PR950: http://llvm.org/PR950 :
Update the test suite to accommodate the change from signed integer types
to signless integer types. The changes were of only a few kinds:

1. Make sure llvm-upgrade is run on the source which does the bulk of the
   changes automatically.

2. Change things like "grep 'int'" to "grep 'i32'"

3. In several tests bitcasting caused the same name to be reused in the
   same type plane. These had to be manually fixed. The fix was (generally)
   to leave the bitcast and provide the instruction with a new name. This
   should not affect the semantics of the test. In a few cases, the 
   bitcasts were known to be superfluous and irrelevant to the test case
   so they were removed.

4. One test case uses a bytecode file which needed to be updated to the
   latest bytecode format.



---
Diffs of the changes:  (+4 -4)

 2005-12-04-DeclarationLineNumbers.c |2 +-
 2006-01-16-BitCountIntrinsicsUnsigned.c |4 ++--
 2006-03-03-MissingInitializer.c |2 +-
 3 files changed, 4 insertions(+), 4 deletions(-)


Index: llvm/test/Regression/CFrontend/2005-12-04-DeclarationLineNumbers.c
diff -u llvm/test/Regression/CFrontend/2005-12-04-DeclarationLineNumbers.c:1.4 
llvm/test/Regression/CFrontend/2005-12-04-DeclarationLineNumbers.c:1.5
--- llvm/test/Regression/CFrontend/2005-12-04-DeclarationLineNumbers.c:1.4  
Fri Jun 16 11:50:24 2006
+++ llvm/test/Regression/CFrontend/2005-12-04-DeclarationLineNumbers.c  Sun Dec 
31 00:01:59 2006
@@ -1,4 +1,4 @@
-// RUN: %llvmgcc %s -S -g -o - | grep 'llvm.dbg.stoppoint.*uint 14'
+// RUN: %llvmgcc %s -S -g -o - | grep 'llvm.dbg.stoppoint.*i32 14'
 // PR664: ensure that line #'s are emitted for declarations
 
 


Index: llvm/test/Regression/CFrontend/2006-01-16-BitCountIntrinsicsUnsigned.c
diff -u 
llvm/test/Regression/CFrontend/2006-01-16-BitCountIntrinsicsUnsigned.c:1.1 
llvm/test/Regression/CFrontend/2006-01-16-BitCountIntrinsicsUnsigned.c:1.2
--- llvm/test/Regression/CFrontend/2006-01-16-BitCountIntrinsicsUnsigned.c:1.1  
Tue Jan 17 00:24:01 2006
+++ llvm/test/Regression/CFrontend/2006-01-16-BitCountIntrinsicsUnsigned.c  
Sun Dec 31 00:01:59 2006
@@ -1,5 +1,5 @@
-// RUN: %llvmgcc -S %s -o - | grep 'llvm.ctlz.i32(uint' &&
-// RUN: %llvmgcc -S %s -o - | not grep 'llvm.ctlz.i32(int'
+// RUN: %llvmgcc -S %s -o - | grep 'llvm.ctlz.i32(i32'
+// RUNMEIFWEHADSIGNEDTYPES: %llvmgcc -S %s -o - | not grep 'llvm.ctlz.i32(i32'
 
 unsigned t2(unsigned X) {
   return __builtin_clz(X);


Index: llvm/test/Regression/CFrontend/2006-03-03-MissingInitializer.c
diff -u llvm/test/Regression/CFrontend/2006-03-03-MissingInitializer.c:1.1 
llvm/test/Regression/CFrontend/2006-03-03-MissingInitializer.c:1.2
--- llvm/test/Regression/CFrontend/2006-03-03-MissingInitializer.c:1.1  Fri Mar 
 3 18:47:12 2006
+++ llvm/test/Regression/CFrontend/2006-03-03-MissingInitializer.c  Sun Dec 
31 00:01:59 2006
@@ -1,4 +1,4 @@
-// RUN: %llvmgcc %s -S -o - | gccas | llvm-dis | grep nate | grep 'global int 
0'
+// RUN: %llvmgcc %s -S -o - | gccas | llvm-dis | grep nate | grep 'global i32 
0'
 
 struct X { int *XX; int Y;};
 



___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

[llvm-commits] CVS: llvm/test/Regression/Transforms/LCSSA/basictest.ll

[Reid Spencer]

Changes in directory llvm/test/Regression/Transforms/LCSSA:

basictest.ll updated: 1.5 -> 1.6
---
Log message:

For PR950: http://llvm.org/PR950 :
Update the test suite to accommodate the change from signed integer types
to signless integer types. The changes were of only a few kinds:

1. Make sure llvm-upgrade is run on the source which does the bulk of the
   changes automatically.

2. Change things like "grep 'int'" to "grep 'i32'"

3. In several tests bitcasting caused the same name to be reused in the
   same type plane. These had to be manually fixed. The fix was (generally)
   to leave the bitcast and provide the instruction with a new name. This
   should not affect the semantics of the test. In a few cases, the 
   bitcasts were known to be superfluous and irrelevant to the test case
   so they were removed.

4. One test case uses a bytecode file which needed to be updated to the
   latest bytecode format.



---
Diffs of the changes:  (+3 -3)

 basictest.ll |6 +++---
 1 files changed, 3 insertions(+), 3 deletions(-)


Index: llvm/test/Regression/Transforms/LCSSA/basictest.ll
diff -u llvm/test/Regression/Transforms/LCSSA/basictest.ll:1.5 
llvm/test/Regression/Transforms/LCSSA/basictest.ll:1.6
--- llvm/test/Regression/Transforms/LCSSA/basictest.ll:1.5  Fri Dec  1 
22:23:09 2006
+++ llvm/test/Regression/Transforms/LCSSA/basictest.ll  Sun Dec 31 00:01:59 2006
@@ -1,5 +1,5 @@
-; RUN: llvm-upgrade < %s | llvm-as | opt -lcssa | llvm-dis | grep "X3.lcssa = 
phi int" &&
-; RUN: llvm-upgrade < %s | llvm-as | opt -lcssa | llvm-dis | grep "%X4 = add 
int 3, %X3.lcssa"
+; RUN: llvm-upgrade < %s | llvm-as | opt -lcssa | llvm-dis | grep "X3.lcssa = 
phi i32" &&
+; RUN: llvm-upgrade < %s | llvm-as | opt -lcssa | llvm-dis | grep "%X4 = add 
i32 3, %X3.lcssa"
 
 void %lcssa(bool %S2) {
 entry:
@@ -23,4 +23,4 @@
 loop.exit:
%X4 = add int 3, %X3
ret void
-}
\ No newline at end of file
+}



___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

[llvm-commits] CVS: llvm/test/Regression/Transforms/LoopStrengthReduce/nested-reduce.ll share_code_in_preheader.ll use_postinc_value_outside_loop.ll var_stride_used_by_compare.ll

[Reid Spencer]

Changes in directory llvm/test/Regression/Transforms/LoopStrengthReduce:

nested-reduce.ll updated: 1.2 -> 1.3
share_code_in_preheader.ll updated: 1.2 -> 1.3
use_postinc_value_outside_loop.ll updated: 1.3 -> 1.4
var_stride_used_by_compare.ll updated: 1.2 -> 1.3
---
Log message:

For PR950: http://llvm.org/PR950 :
Update the test suite to accommodate the change from signed integer types
to signless integer types. The changes were of only a few kinds:

1. Make sure llvm-upgrade is run on the source which does the bulk of the
   changes automatically.

2. Change things like "grep 'int'" to "grep 'i32'"

3. In several tests bitcasting caused the same name to be reused in the
   same type plane. These had to be manually fixed. The fix was (generally)
   to leave the bitcast and provide the instruction with a new name. This
   should not affect the semantics of the test. In a few cases, the 
   bitcasts were known to be superfluous and irrelevant to the test case
   so they were removed.

4. One test case uses a bytecode file which needed to be updated to the
   latest bytecode format.



---
Diffs of the changes:  (+18 -18)

 nested-reduce.ll  |6 +++---
 share_code_in_preheader.ll|   12 ++--
 use_postinc_value_outside_loop.ll |   10 +-
 var_stride_used_by_compare.ll |8 
 4 files changed, 18 insertions(+), 18 deletions(-)


Index: llvm/test/Regression/Transforms/LoopStrengthReduce/nested-reduce.ll
diff -u llvm/test/Regression/Transforms/LoopStrengthReduce/nested-reduce.ll:1.2 
llvm/test/Regression/Transforms/LoopStrengthReduce/nested-reduce.ll:1.3
--- llvm/test/Regression/Transforms/LoopStrengthReduce/nested-reduce.ll:1.2 
Fri Dec  1 22:23:09 2006
+++ llvm/test/Regression/Transforms/LoopStrengthReduce/nested-reduce.ll Sun Dec 
31 00:01:59 2006
@@ -25,8 +25,8 @@
br label %no_exit.1
 
 no_exit.1: ; preds = %cond_continue, %no_exit.1.outer
-   %indvar = phi uint [ 0, %no_exit.1.outer ], [ %indvar.next, 
%cond_continue ];  [#uses=2]
-   %indvar = cast uint %indvar to int  ;  [#uses=1]
+   %indvar.ui = phi uint [ 0, %no_exit.1.outer ], [ %indvar.next, 
%cond_continue ] ;  [#uses=2]
+   %indvar = cast uint %indvar.ui to int   ;  [#uses=1]
%j.1.2 = add int %indvar, %j.1.2.ph ;  [#uses=2]
%tmp.11 = add int %j.1.2, %tmp.9;  [#uses=1]
%tmp.12 = cast int %tmp.11 to ubyte ;  [#uses=1]
@@ -43,7 +43,7 @@
 
 cond_continue: ; preds = %no_exit.1
%tmp.519 = setlt int %inc.1, %C ;  [#uses=1]
-   %indvar.next = add uint %indvar, 1  ;  [#uses=1]
+   %indvar.next = add uint %indvar.ui, 1   ;  [#uses=1]
br bool %tmp.519, label %no_exit.1, label %loopexit.1
 
 loopexit.1:; preds = %cond_continue, %cond_true, %loopentry.1


Index: 
llvm/test/Regression/Transforms/LoopStrengthReduce/share_code_in_preheader.ll
diff -u 
llvm/test/Regression/Transforms/LoopStrengthReduce/share_code_in_preheader.ll:1.2
 
llvm/test/Regression/Transforms/LoopStrengthReduce/share_code_in_preheader.ll:1.3
--- 
llvm/test/Regression/Transforms/LoopStrengthReduce/share_code_in_preheader.ll:1.2
   Fri Dec  1 22:23:09 2006
+++ 
llvm/test/Regression/Transforms/LoopStrengthReduce/share_code_in_preheader.ll   
Sun Dec 31 00:01:59 2006
@@ -1,17 +1,17 @@
 ; RUN: llvm-upgrade < %s | llvm-as | opt -loop-reduce | llvm-dis | grep mul | 
wc -l | grep 1
 ; LSR should not make two copies of the Q*L expression in the preheader!
 
-sbyte %test(sbyte* %A, sbyte* %B, int %L, int %Q, int %N) {
+sbyte %test(sbyte* %A, sbyte* %B, int %L, int %Q, int %N.s) {
 entry:
%tmp.6 = mul int %Q, %L ;  [#uses=1]
-   %N = cast int %N to uint;  [#uses=1]
+   %N = cast int %N.s to uint  ;  [#uses=1]
br label %no_exit
 
 no_exit:   ; preds = %no_exit, %no_exit.preheader
-   %indvar = phi uint [ 0, %entry], [ %indvar.next, %no_exit ] 
;  [#uses=2]
+   %indvar.ui = phi uint [ 0, %entry], [ %indvar.next, %no_exit ]  
;  [#uses=2]
%Sum.0.0 = phi sbyte [ 0, %entry], [ %tmp.21, %no_exit ]
;  [#uses=1]
-   %indvar = cast uint %indvar to int  ;  [#uses=1]
-   %N_addr.0.0 = sub int %N, %indvar   ;  [#uses=1]
+   %indvar = cast uint %indvar.ui to int   ;  [#uses=1]
+   %N_addr.0.0 = sub int %N.s, %indvar ;  [#uses=1]
%tmp.8 = add int %N_addr.0.0, %tmp.6;  [#uses=2]
%tmp.9 = getelementptr sbyte* %A, int %tmp.8;  
[#uses=1]
%tmp.10 = load sbyte* %tmp.9;  [#uses=1]
@@ -19,7 +19,7 @@
%tmp.18 = load sbyte* %tmp.17   ;  [#uses=1]
%tmp.19 = sub sbyte %tmp.10, %tmp.18;  [#uses=1]
%tmp.21 = add sbyte %tmp.19, %Sum.0.0   ;  [#uses=2]
-   %indvar.nex

[llvm-commits] CVS: llvm/test/Regression/CodeGen/Generic/2005-10-21-longlonggtu.ll 2006-06-28-SimplifySetCCCrash.ll SwitchLowering.ll

[Reid Spencer]

Changes in directory llvm/test/Regression/CodeGen/Generic:

2005-10-21-longlonggtu.ll updated: 1.2 -> 1.3
2006-06-28-SimplifySetCCCrash.ll updated: 1.2 -> 1.3
SwitchLowering.ll updated: 1.2 -> 1.3
---
Log message:

For PR950: http://llvm.org/PR950 :
Update the test suite to accommodate the change from signed integer types
to signless integer types. The changes were of only a few kinds:

1. Make sure llvm-upgrade is run on the source which does the bulk of the
   changes automatically.

2. Change things like "grep 'int'" to "grep 'i32'"

3. In several tests bitcasting caused the same name to be reused in the
   same type plane. These had to be manually fixed. The fix was (generally)
   to leave the bitcast and provide the instruction with a new name. This
   should not affect the semantics of the test. In a few cases, the 
   bitcasts were known to be superfluous and irrelevant to the test case
   so they were removed.

4. One test case uses a bytecode file which needed to be updated to the
   latest bytecode format.



---
Diffs of the changes:  (+7 -8)

 2005-10-21-longlonggtu.ll|4 ++--
 2006-06-28-SimplifySetCCCrash.ll |   10 +-
 SwitchLowering.ll|1 -
 3 files changed, 7 insertions(+), 8 deletions(-)


Index: llvm/test/Regression/CodeGen/Generic/2005-10-21-longlonggtu.ll
diff -u llvm/test/Regression/CodeGen/Generic/2005-10-21-longlonggtu.ll:1.2 
llvm/test/Regression/CodeGen/Generic/2005-10-21-longlonggtu.ll:1.3
--- llvm/test/Regression/CodeGen/Generic/2005-10-21-longlonggtu.ll:1.2  Fri Dec 
 1 22:23:08 2006
+++ llvm/test/Regression/CodeGen/Generic/2005-10-21-longlonggtu.ll  Sun Dec 
31 00:01:59 2006
@@ -1,6 +1,6 @@
 ; RUN: llvm-upgrade < %s | llvm-as | llc
-float %t(long %u) {
-   %u = cast long %u to ulong  ;  [#uses=1]
+float %t(long %u_arg) {
+   %u = cast long %u_arg to ulong  ;  [#uses=1]
%tmp5 = add ulong %u, 9007199254740991  ;  [#uses=1]
%tmp = setgt ulong %tmp5, 18014398509481982 ;  
[#uses=1]
br bool %tmp, label %T, label %F


Index: llvm/test/Regression/CodeGen/Generic/2006-06-28-SimplifySetCCCrash.ll
diff -u 
llvm/test/Regression/CodeGen/Generic/2006-06-28-SimplifySetCCCrash.ll:1.2 
llvm/test/Regression/CodeGen/Generic/2006-06-28-SimplifySetCCCrash.ll:1.3
--- llvm/test/Regression/CodeGen/Generic/2006-06-28-SimplifySetCCCrash.ll:1.2   
Fri Dec  1 22:23:08 2006
+++ llvm/test/Regression/CodeGen/Generic/2006-06-28-SimplifySetCCCrash.ll   
Sun Dec 31 00:01:59 2006
@@ -239,19 +239,19 @@
ret void
 
 bb1876:; preds = %bb1808
-   %tmp1877 = load int* %which_alternative ;  [#uses=4]
-   %tmp1877 = cast int %tmp1877 to uint;  [#uses=1]
+   %tmp1877signed = load int* %which_alternative   ;  [#uses=4]
+   %tmp1877 = cast int %tmp1877signed to uint  ;  [#uses=1]
%bothcond699 = setlt uint %tmp1877, 2   ;  [#uses=1]
-   %tmp1888 = seteq int %tmp1877, 2;  [#uses=1]
+   %tmp1888 = seteq int %tmp1877signed, 2  ;  [#uses=1]
%bothcond700 = or bool %bothcond699, %tmp1888   ;  
[#uses=1]
%bothcond700.not = xor bool %bothcond700, true  ;  
[#uses=1]
-   %tmp1894 = seteq int %tmp1877, 3;  [#uses=1]
+   %tmp1894 = seteq int %tmp1877signed, 3  ;  [#uses=1]
%bothcond701 = or bool %tmp1894, %bothcond700.not   ; 
 [#uses=1]
%bothcond702 = or bool %bothcond701, false  ;  
[#uses=1]
br bool %bothcond702, label %UnifiedReturnBlock, label %cond_next1902
 
 cond_next1902: ; preds = %bb1876
-   switch int %tmp1877, label %cond_next1937 [
+   switch int %tmp1877signed, label %cond_next1937 [
 int 0, label %bb1918
 int 1, label %bb1918
 int 2, label %bb1918


Index: llvm/test/Regression/CodeGen/Generic/SwitchLowering.ll
diff -u llvm/test/Regression/CodeGen/Generic/SwitchLowering.ll:1.2 
llvm/test/Regression/CodeGen/Generic/SwitchLowering.ll:1.3
--- llvm/test/Regression/CodeGen/Generic/SwitchLowering.ll:1.2  Fri Dec  1 
22:23:08 2006
+++ llvm/test/Regression/CodeGen/Generic/SwitchLowering.ll  Sun Dec 31 
00:01:59 2006
@@ -19,7 +19,6 @@
]
 
 bb7:   ; preds = %bb, %bb
-   %tmp = cast sbyte %tmp to ubyte ;  [#uses=1]
tail call void %foo( ubyte %tmp )
ret sbyte* %tmp2
 }



___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

[llvm-commits] CVS: llvm/test/Regression/Transforms/SCCP/2004-12-10-UndefBranchBug.ll 2006-12-04-PackedType.ll ipsccp-basic.ll ipsccp-conditional.ll logical-nuke.ll

[Reid Spencer]

Changes in directory llvm/test/Regression/Transforms/SCCP:

2004-12-10-UndefBranchBug.ll updated: 1.2 -> 1.3
2006-12-04-PackedType.ll updated: 1.1 -> 1.2
ipsccp-basic.ll updated: 1.3 -> 1.4
ipsccp-conditional.ll updated: 1.3 -> 1.4
logical-nuke.ll updated: 1.2 -> 1.3
---
Log message:

For PR950: http://llvm.org/PR950 :
Update the test suite to accommodate the change from signed integer types
to signless integer types. The changes were of only a few kinds:

1. Make sure llvm-upgrade is run on the source which does the bulk of the
   changes automatically.

2. Change things like "grep 'int'" to "grep 'i32'"

3. In several tests bitcasting caused the same name to be reused in the
   same type plane. These had to be manually fixed. The fix was (generally)
   to leave the bitcast and provide the instruction with a new name. This
   should not affect the semantics of the test. In a few cases, the 
   bitcasts were known to be superfluous and irrelevant to the test case
   so they were removed.

4. One test case uses a bytecode file which needed to be updated to the
   latest bytecode format.



---
Diffs of the changes:  (+6 -6)

 2004-12-10-UndefBranchBug.ll |2 +-
 2006-12-04-PackedType.ll |4 ++--
 ipsccp-basic.ll  |2 +-
 ipsccp-conditional.ll|2 +-
 logical-nuke.ll  |2 +-
 5 files changed, 6 insertions(+), 6 deletions(-)


Index: llvm/test/Regression/Transforms/SCCP/2004-12-10-UndefBranchBug.ll
diff -u llvm/test/Regression/Transforms/SCCP/2004-12-10-UndefBranchBug.ll:1.2 
llvm/test/Regression/Transforms/SCCP/2004-12-10-UndefBranchBug.ll:1.3
--- llvm/test/Regression/Transforms/SCCP/2004-12-10-UndefBranchBug.ll:1.2   
Fri Dec  1 22:23:10 2006
+++ llvm/test/Regression/Transforms/SCCP/2004-12-10-UndefBranchBug.ll   Sun Dec 
31 00:02:00 2006
@@ -1,4 +1,4 @@
-; RUN: llvm-upgrade < %s | llvm-as | opt -sccp | llvm-dis | grep 'ret int 1'
+; RUN: llvm-upgrade < %s | llvm-as | opt -sccp | llvm-dis | grep 'ret i32 1'
 
 ; This function definitely returns 1, even if we don't know the direction
 ; of the branch.


Index: llvm/test/Regression/Transforms/SCCP/2006-12-04-PackedType.ll
diff -u llvm/test/Regression/Transforms/SCCP/2006-12-04-PackedType.ll:1.1 
llvm/test/Regression/Transforms/SCCP/2006-12-04-PackedType.ll:1.2
--- llvm/test/Regression/Transforms/SCCP/2006-12-04-PackedType.ll:1.1   Mon Dec 
 4 17:54:59 2006
+++ llvm/test/Regression/Transforms/SCCP/2006-12-04-PackedType.ll   Sun Dec 
31 00:02:00 2006
@@ -105,8 +105,8 @@
 
 void %gldLLVMVecPointRender(%struct.GLDContextRec* %ctx) {
 entry:
-   %tmp = getelementptr %struct.GLDContextRec* %ctx, int 0, uint 22
;  [#uses=1]
-   %tmp = load uint* %tmp  ;  [#uses=3]
+   %tmp.uip = getelementptr %struct.GLDContextRec* %ctx, int 0, uint 22
;  [#uses=1]
+   %tmp = load uint* %tmp.uip  ;  [#uses=3]
%tmp91 = lshr uint %tmp, ubyte 5;  [#uses=1]
%tmp92 = trunc uint %tmp91 to bool  ;  [#uses=1]
br bool %tmp92, label %cond_true93, label %cond_next116


Index: llvm/test/Regression/Transforms/SCCP/ipsccp-basic.ll
diff -u llvm/test/Regression/Transforms/SCCP/ipsccp-basic.ll:1.3 
llvm/test/Regression/Transforms/SCCP/ipsccp-basic.ll:1.4
--- llvm/test/Regression/Transforms/SCCP/ipsccp-basic.ll:1.3Fri Dec  1 
22:23:10 2006
+++ llvm/test/Regression/Transforms/SCCP/ipsccp-basic.llSun Dec 31 
00:02:00 2006
@@ -1,4 +1,4 @@
-; RUN: llvm-upgrade < %s | llvm-as | opt -ipsccp | llvm-dis | grep -v 'ret int 
17' | grep -v 'ret int undef' | not grep ret
+; RUN: llvm-upgrade < %s | llvm-as | opt -ipsccp | llvm-dis | grep -v 'ret i32 
17' | grep -v 'ret i32 undef' | not grep ret
 
 implementation
 


Index: llvm/test/Regression/Transforms/SCCP/ipsccp-conditional.ll
diff -u llvm/test/Regression/Transforms/SCCP/ipsccp-conditional.ll:1.3 
llvm/test/Regression/Transforms/SCCP/ipsccp-conditional.ll:1.4
--- llvm/test/Regression/Transforms/SCCP/ipsccp-conditional.ll:1.3  Fri Dec 
 1 22:23:10 2006
+++ llvm/test/Regression/Transforms/SCCP/ipsccp-conditional.ll  Sun Dec 31 
00:02:00 2006
@@ -1,4 +1,4 @@
-; RUN: llvm-upgrade < %s | llvm-as | opt -ipsccp | llvm-dis | grep -v 'ret int 
0' | grep -v 'ret int undef' | not grep ret
+; RUN: llvm-upgrade < %s | llvm-as | opt -ipsccp | llvm-dis | grep -v 'ret i32 
0' | grep -v 'ret i32 undef' | not grep ret
 
 implementation
 


Index: llvm/test/Regression/Transforms/SCCP/logical-nuke.ll
diff -u llvm/test/Regression/Transforms/SCCP/logical-nuke.ll:1.2 
llvm/test/Regression/Transforms/SCCP/logical-nuke.ll:1.3
--- llvm/test/Regression/Transforms/SCCP/logical-nuke.ll:1.2Fri Dec  1 
22:23:10 2006
+++ llvm/test/Regression/Transforms/SCCP/logical-nuke.llSun Dec 31 
00:02:00 2006
@@ -1,4 +1,4 @@
-; RUN: llvm-upgrade < %s | llvm-as | opt -sccp | llvm-dis | grep 'ret int 0'
+; RUN: llvm-upgrade < %s | llvm-as | opt -sccp | llvm-dis | grep 'ret i32 0'
 
 ; Test that SCCP has

[llvm-commits] CVS: llvm/test/Regression/Transforms/PredicateSimplifier/2006-09-20-SetCC.ll

[Reid Spencer]

Changes in directory llvm/test/Regression/Transforms/PredicateSimplifier:

2006-09-20-SetCC.ll updated: 1.3 -> 1.4
---
Log message:

For PR950: http://llvm.org/PR950 :
Update the test suite to accommodate the change from signed integer types
to signless integer types. The changes were of only a few kinds:

1. Make sure llvm-upgrade is run on the source which does the bulk of the
   changes automatically.

2. Change things like "grep 'int'" to "grep 'i32'"

3. In several tests bitcasting caused the same name to be reused in the
   same type plane. These had to be manually fixed. The fix was (generally)
   to leave the bitcast and provide the instruction with a new name. This
   should not affect the semantics of the test. In a few cases, the 
   bitcasts were known to be superfluous and irrelevant to the test case
   so they were removed.

4. One test case uses a bytecode file which needed to be updated to the
   latest bytecode format.



---
Diffs of the changes:  (+3 -3)

 2006-09-20-SetCC.ll |6 +++---
 1 files changed, 3 insertions(+), 3 deletions(-)


Index: llvm/test/Regression/Transforms/PredicateSimplifier/2006-09-20-SetCC.ll
diff -u 
llvm/test/Regression/Transforms/PredicateSimplifier/2006-09-20-SetCC.ll:1.3 
llvm/test/Regression/Transforms/PredicateSimplifier/2006-09-20-SetCC.ll:1.4
--- llvm/test/Regression/Transforms/PredicateSimplifier/2006-09-20-SetCC.ll:1.3 
Fri Dec  1 22:23:10 2006
+++ llvm/test/Regression/Transforms/PredicateSimplifier/2006-09-20-SetCC.ll 
Sun Dec 31 00:01:59 2006
@@ -16,8 +16,8 @@
br label %cond_true
 
 cond_true: ; preds = %return.i, %cond_true.outer
-   %indvar = phi uint [ 0, %cond_true.outer ], [ %indvar.next, %return.i ] 
;  [#uses=2]
-   %indvar = cast uint %indvar to int  ;  [#uses=1]
+   %indvar.ui = phi uint [ 0, %cond_true.outer ], [ %indvar.next, 
%return.i ]  ;  [#uses=2]
+   %indvar = cast uint %indvar.ui to int   ;  [#uses=1]
%i.0.0 = add int %indvar, %i.0.0.ph ;  [#uses=3]
%savedstack = call sbyte* %llvm.stacksave( );  
[#uses=2]
%tmp.i = seteq int %i.0.0, 0;  [#uses=1]
@@ -38,7 +38,7 @@
 return.i:  ; preds = %cond_true
call void %llvm.stackrestore( sbyte* %savedstack )
%tmp21 = setgt int %tmp5,   ;  [#uses=1]
-   %indvar.next = add uint %indvar, 1  ;  [#uses=1]
+   %indvar.next = add uint %indvar.ui, 1   ;  [#uses=1]
br bool %tmp21, label %bb8, label %cond_true
 
 bb8:   ; preds = %return.i, %cond_true.i



___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

[llvm-commits] CVS: llvm/test/Regression/Analysis/GlobalsModRef/indirect-global.ll

[Reid Spencer]

Changes in directory llvm/test/Regression/Analysis/GlobalsModRef:

indirect-global.ll updated: 1.2 -> 1.3
---
Log message:

For PR950: http://llvm.org/PR950 :
Update the test suite to accommodate the change from signed integer types
to signless integer types. The changes were of only a few kinds:

1. Make sure llvm-upgrade is run on the source which does the bulk of the
   changes automatically.

2. Change things like "grep 'int'" to "grep 'i32'"

3. In several tests bitcasting caused the same name to be reused in the
   same type plane. These had to be manually fixed. The fix was (generally)
   to leave the bitcast and provide the instruction with a new name. This
   should not affect the semantics of the test. In a few cases, the 
   bitcasts were known to be superfluous and irrelevant to the test case
   so they were removed.

4. One test case uses a bytecode file which needed to be updated to the
   latest bytecode format.



---
Diffs of the changes:  (+3 -1)

 indirect-global.ll |4 +++-
 1 files changed, 3 insertions(+), 1 deletion(-)


Index: llvm/test/Regression/Analysis/GlobalsModRef/indirect-global.ll
diff -u llvm/test/Regression/Analysis/GlobalsModRef/indirect-global.ll:1.2 
llvm/test/Regression/Analysis/GlobalsModRef/indirect-global.ll:1.3
--- llvm/test/Regression/Analysis/GlobalsModRef/indirect-global.ll:1.2  Fri Dec 
 1 22:23:07 2006
+++ llvm/test/Regression/Analysis/GlobalsModRef/indirect-global.ll  Sun Dec 
31 00:01:59 2006
@@ -1,4 +1,6 @@
-; RUN: llvm-upgrade < %s | llvm-as | opt -globalsmodref-aa -load-vn -gcse 
-instcombine | llvm-dis | grep 'ret int 0'
+; RUN: llvm-upgrade < %s | llvm-as | \
+; RUN:   opt -globalsmodref-aa -load-vn -gcse -instcombine | llvm-dis | \
+; RUN:   grep 'ret i32 0'
 %G = internal global int* null
 
 implementation



___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

[llvm-commits] CVS: llvm/test/Regression/Transforms/TailCallElim/dont-tce-tail-marked-call.ll move_alloca_for_tail_call.ll

[Reid Spencer]

Changes in directory llvm/test/Regression/Transforms/TailCallElim:

dont-tce-tail-marked-call.ll updated: 1.2 -> 1.3
move_alloca_for_tail_call.ll updated: 1.2 -> 1.3
---
Log message:

For PR950: http://llvm.org/PR950 :
Update the test suite to accommodate the change from signed integer types
to signless integer types. The changes were of only a few kinds:

1. Make sure llvm-upgrade is run on the source which does the bulk of the
   changes automatically.

2. Change things like "grep 'int'" to "grep 'i32'"

3. In several tests bitcasting caused the same name to be reused in the
   same type plane. These had to be manually fixed. The fix was (generally)
   to leave the bitcast and provide the instruction with a new name. This
   should not affect the semantics of the test. In a few cases, the 
   bitcasts were known to be superfluous and irrelevant to the test case
   so they were removed.

4. One test case uses a bytecode file which needed to be updated to the
   latest bytecode format.



---
Diffs of the changes:  (+2 -2)

 dont-tce-tail-marked-call.ll |2 +-
 move_alloca_for_tail_call.ll |2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)


Index: llvm/test/Regression/Transforms/TailCallElim/dont-tce-tail-marked-call.ll
diff -u 
llvm/test/Regression/Transforms/TailCallElim/dont-tce-tail-marked-call.ll:1.2 
llvm/test/Regression/Transforms/TailCallElim/dont-tce-tail-marked-call.ll:1.3
--- 
llvm/test/Regression/Transforms/TailCallElim/dont-tce-tail-marked-call.ll:1.2   
Fri Dec  1 22:23:10 2006
+++ llvm/test/Regression/Transforms/TailCallElim/dont-tce-tail-marked-call.ll   
Sun Dec 31 00:02:00 2006
@@ -1,4 +1,4 @@
-; RUN: llvm-upgrade < %s | llvm-as | opt -tailcallelim | llvm-dis | grep 'call 
int %foo'
+; RUN: llvm-upgrade < %s | llvm-as | opt -tailcallelim | llvm-dis | grep 'call 
i32 %foo'
 
 declare void %bar(int*)
 int %foo(uint %N) {


Index: llvm/test/Regression/Transforms/TailCallElim/move_alloca_for_tail_call.ll
diff -u 
llvm/test/Regression/Transforms/TailCallElim/move_alloca_for_tail_call.ll:1.2 
llvm/test/Regression/Transforms/TailCallElim/move_alloca_for_tail_call.ll:1.3
--- 
llvm/test/Regression/Transforms/TailCallElim/move_alloca_for_tail_call.ll:1.2   
Fri Dec  1 22:23:10 2006
+++ llvm/test/Regression/Transforms/TailCallElim/move_alloca_for_tail_call.ll   
Sun Dec 31 00:02:00 2006
@@ -1,4 +1,4 @@
-; RUN: llvm-upgrade < %s | llvm-as | opt -tailcallelim | llvm-dis | %prcontext 
alloca 1 | grep 'int %foo'
+; RUN: llvm-upgrade < %s | llvm-as | opt -tailcallelim | llvm-dis | %prcontext 
alloca 1 | grep 'i32 %foo'
 
 declare void %bar(int*)
 int %foo() {



___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

[llvm-commits] CVS: llvm/test/Regression/Transforms/SimplifyCFG/2006-10-29-InvokeCrash.ll 2006-12-08-Ptr-ICmp-Branch.ll

[Reid Spencer]

Changes in directory llvm/test/Regression/Transforms/SimplifyCFG:

2006-10-29-InvokeCrash.ll updated: 1.2 -> 1.3
2006-12-08-Ptr-ICmp-Branch.ll updated: 1.2 -> 1.3
---
Log message:

For PR950: http://llvm.org/PR950 :
Update the test suite to accommodate the change from signed integer types
to signless integer types. The changes were of only a few kinds:

1. Make sure llvm-upgrade is run on the source which does the bulk of the
   changes automatically.

2. Change things like "grep 'int'" to "grep 'i32'"

3. In several tests bitcasting caused the same name to be reused in the
   same type plane. These had to be manually fixed. The fix was (generally)
   to leave the bitcast and provide the instruction with a new name. This
   should not affect the semantics of the test. In a few cases, the 
   bitcasts were known to be superfluous and irrelevant to the test case
   so they were removed.

4. One test case uses a bytecode file which needed to be updated to the
   latest bytecode format.



---
Diffs of the changes:  (+3 -3)

 2006-10-29-InvokeCrash.ll |4 ++--
 2006-12-08-Ptr-ICmp-Branch.ll |2 +-
 2 files changed, 3 insertions(+), 3 deletions(-)


Index: llvm/test/Regression/Transforms/SimplifyCFG/2006-10-29-InvokeCrash.ll
diff -u 
llvm/test/Regression/Transforms/SimplifyCFG/2006-10-29-InvokeCrash.ll:1.2 
llvm/test/Regression/Transforms/SimplifyCFG/2006-10-29-InvokeCrash.ll:1.3
--- llvm/test/Regression/Transforms/SimplifyCFG/2006-10-29-InvokeCrash.ll:1.2   
Fri Dec  1 22:23:10 2006
+++ llvm/test/Regression/Transforms/SimplifyCFG/2006-10-29-InvokeCrash.ll   
Sun Dec 31 00:02:00 2006
@@ -413,11 +413,11 @@
 cond_next225:  ; preds = %invcont222, %cond_true217, %invcont213
%toPage.1 = phi int [ %tmp223, %invcont222 ], [ %tmp214, %cond_true217 
], [ %tmp214, %invcont213 ]  ;  [#uses=2]
%fromPage.1 = phi int [ 1, %invcont222 ], [ %tmp211, %cond_true217 ], [ 
%tmp211, %invcont213 ]  ;  [#uses=2]
-   %tmp = invoke uint %_ZNK8QPrinter9pageOrderEv( %struct.QPrinter* 
%printer )
+   %tmp.page = invoke uint %_ZNK8QPrinter9pageOrderEv( %struct.QPrinter* 
%printer )
to label %invcont227 unwind label %cleanup329   
;  [#uses=1]
 
 invcont227:; preds = %cond_next225
-   %tmp228 = seteq uint %tmp, 1;  [#uses=1]
+   %tmp228 = seteq uint %tmp.page, 1   ;  [#uses=1]
br bool %tmp228, label %cond_true230, label %cond_next234
 
 cond_true230:  ; preds = %invcont227


Index: llvm/test/Regression/Transforms/SimplifyCFG/2006-12-08-Ptr-ICmp-Branch.ll
diff -u 
llvm/test/Regression/Transforms/SimplifyCFG/2006-12-08-Ptr-ICmp-Branch.ll:1.2 
llvm/test/Regression/Transforms/SimplifyCFG/2006-12-08-Ptr-ICmp-Branch.ll:1.3
--- 
llvm/test/Regression/Transforms/SimplifyCFG/2006-12-08-Ptr-ICmp-Branch.ll:1.2   
Fri Dec 29 14:01:32 2006
+++ llvm/test/Regression/Transforms/SimplifyCFG/2006-12-08-Ptr-ICmp-Branch.ll   
Sun Dec 31 00:02:00 2006
@@ -56,7 +56,7 @@
 
 bb: ; preds = %bb33
 %tmp = load %struct.FILE** %f_addr  ; <%struct.FILE*> 
[#uses=1]
-%tmp = call int %_IO_getc( %struct.FILE* %tmp ) ;  
[#uses=1]
+%tmp.r = call int %_IO_getc( %struct.FILE* %tmp ) ;  
[#uses=1]
 %tmp6 = call int %tolower( int %tmp )   ;  [#uses=1]
 %tmp6 = trunc int %tmp6 to sbyte;  [#uses=1]
 store sbyte %tmp6, sbyte* %c



___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

[llvm-commits] CVS: llvm/test/Regression/Analysis/BasicAA/2004-07-28-MustAliasbug.llx 2005-03-09-BrokenBasicAA.ll tailcall-modref.ll

[Reid Spencer]

Changes in directory llvm/test/Regression/Analysis/BasicAA:

2004-07-28-MustAliasbug.llx updated: 1.2 -> 1.3
2005-03-09-BrokenBasicAA.ll updated: 1.2 -> 1.3
tailcall-modref.ll updated: 1.2 -> 1.3
---
Log message:

For PR950: http://llvm.org/PR950 :
Update the test suite to accommodate the change from signed integer types
to signless integer types. The changes were of only a few kinds:

1. Make sure llvm-upgrade is run on the source which does the bulk of the
   changes automatically.

2. Change things like "grep 'int'" to "grep 'i32'"

3. In several tests bitcasting caused the same name to be reused in the
   same type plane. These had to be manually fixed. The fix was (generally)
   to leave the bitcast and provide the instruction with a new name. This
   should not affect the semantics of the test. In a few cases, the 
   bitcasts were known to be superfluous and irrelevant to the test case
   so they were removed.

4. One test case uses a bytecode file which needed to be updated to the
   latest bytecode format.



---
Diffs of the changes:  (+3 -3)

 2004-07-28-MustAliasbug.llx |2 +-
 2005-03-09-BrokenBasicAA.ll |2 +-
 tailcall-modref.ll  |2 +-
 3 files changed, 3 insertions(+), 3 deletions(-)


Index: llvm/test/Regression/Analysis/BasicAA/2004-07-28-MustAliasbug.llx
diff -u llvm/test/Regression/Analysis/BasicAA/2004-07-28-MustAliasbug.llx:1.2 
llvm/test/Regression/Analysis/BasicAA/2004-07-28-MustAliasbug.llx:1.3
--- llvm/test/Regression/Analysis/BasicAA/2004-07-28-MustAliasbug.llx:1.2   
Fri Dec 29 14:01:32 2006
+++ llvm/test/Regression/Analysis/BasicAA/2004-07-28-MustAliasbug.llx   Sun Dec 
31 00:01:59 2006
@@ -1,4 +1,4 @@
-; RUN: llvm-upgrade < %s | llvm-as | opt -dse | llvm-dis | grep 'store int 0'
+; RUN: llvm-upgrade < %s | llvm-as | opt -dse | llvm-dis | grep 'store i32 0'
 
 void %test({int,int }* %P) {
%Q = getelementptr {int,int}* %P, int 1


Index: llvm/test/Regression/Analysis/BasicAA/2005-03-09-BrokenBasicAA.ll
diff -u llvm/test/Regression/Analysis/BasicAA/2005-03-09-BrokenBasicAA.ll:1.2 
llvm/test/Regression/Analysis/BasicAA/2005-03-09-BrokenBasicAA.ll:1.3
--- llvm/test/Regression/Analysis/BasicAA/2005-03-09-BrokenBasicAA.ll:1.2   
Fri Dec  1 22:23:07 2006
+++ llvm/test/Regression/Analysis/BasicAA/2005-03-09-BrokenBasicAA.ll   Sun Dec 
31 00:01:59 2006
@@ -1,4 +1,4 @@
-; RUN: llvm-upgrade < %s | llvm-as | opt -basicaa -load-vn -gcse -instcombine 
| llvm-dis | grep 'load int\* %A'
+; RUN: llvm-upgrade < %s | llvm-as | opt -basicaa -load-vn -gcse -instcombine 
| llvm-dis | grep 'load i32\* %A'
 
 declare double* %useit(int*)
 


Index: llvm/test/Regression/Analysis/BasicAA/tailcall-modref.ll
diff -u llvm/test/Regression/Analysis/BasicAA/tailcall-modref.ll:1.2 
llvm/test/Regression/Analysis/BasicAA/tailcall-modref.ll:1.3
--- llvm/test/Regression/Analysis/BasicAA/tailcall-modref.ll:1.2Fri Dec 
 1 22:23:07 2006
+++ llvm/test/Regression/Analysis/BasicAA/tailcall-modref.llSun Dec 31 
00:01:59 2006
@@ -1,4 +1,4 @@
-; RUN: llvm-upgrade < %s | llvm-as | opt -basicaa -load-vn -gcse -instcombine 
| llvm-dis | grep 'ret int 0'
+; RUN: llvm-upgrade < %s | llvm-as | opt -basicaa -load-vn -gcse -instcombine 
| llvm-dis | grep 'ret i32 0'
 declare void %foo(int*)
 declare void %bar()
 



___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

[llvm-commits] CVS: llvm/test/Regression/ExecutionEngine/2003-01-04-PhiTest.ll 2003-05-11-PHIRegAllocBug.ll 2003-06-05-PHIBug.ll 2003-08-15-AllocaAssertion.ll 2003-08-23-RegisterAllocatePhysReg.ll hel

[Reid Spencer]

Changes in directory llvm/test/Regression/ExecutionEngine:

2003-01-04-PhiTest.ll updated: 1.4 -> 1.5
2003-05-11-PHIRegAllocBug.ll updated: 1.3 -> 1.4
2003-06-05-PHIBug.ll updated: 1.2 -> 1.3
2003-08-15-AllocaAssertion.ll updated: 1.2 -> 1.3
2003-08-23-RegisterAllocatePhysReg.ll updated: 1.2 -> 1.3
hello.ll updated: 1.3 -> 1.4
hello2.ll updated: 1.3 -> 1.4
simplesttest.ll updated: 1.5 -> 1.6
simpletest.ll updated: 1.2 -> 1.3
test-loadstore.ll updated: 1.6 -> 1.7
test-logical.ll updated: 1.5 -> 1.6
test-malloc.ll updated: 1.2 -> 1.3
test-ret.ll updated: 1.5 -> 1.6
test-shift.ll updated: 1.7 -> 1.8
---
Log message:

For PR950: http://llvm.org/PR950 :
Update the test suite to accommodate the change from signed integer types
to signless integer types. The changes were of only a few kinds:

1. Make sure llvm-upgrade is run on the source which does the bulk of the
   changes automatically.

2. Change things like "grep 'int'" to "grep 'i32'"

3. In several tests bitcasting caused the same name to be reused in the
   same type plane. These had to be manually fixed. The fix was (generally)
   to leave the bitcast and provide the instruction with a new name. This
   should not affect the semantics of the test. In a few cases, the 
   bitcasts were known to be superfluous and irrelevant to the test case
   so they were removed.

4. One test case uses a bytecode file which needed to be updated to the
   latest bytecode format.



---
Diffs of the changes:  (+36 -36)

 2003-01-04-PhiTest.ll |2 +-
 2003-05-11-PHIRegAllocBug.ll  |2 +-
 2003-06-05-PHIBug.ll  |2 +-
 2003-08-15-AllocaAssertion.ll |2 +-
 2003-08-23-RegisterAllocatePhysReg.ll |2 +-
 hello.ll  |2 +-
 hello2.ll |2 +-
 simplesttest.ll   |2 +-
 simpletest.ll |2 +-
 test-loadstore.ll |2 +-
 test-logical.ll   |2 +-
 test-malloc.ll|2 +-
 test-ret.ll   |   22 +++---
 test-shift.ll |   26 +-
 14 files changed, 36 insertions(+), 36 deletions(-)


Index: llvm/test/Regression/ExecutionEngine/2003-01-04-PhiTest.ll
diff -u llvm/test/Regression/ExecutionEngine/2003-01-04-PhiTest.ll:1.4 
llvm/test/Regression/ExecutionEngine/2003-01-04-PhiTest.ll:1.5
--- llvm/test/Regression/ExecutionEngine/2003-01-04-PhiTest.ll:1.4  Sat Nov 
 6 17:32:43 2004
+++ llvm/test/Regression/ExecutionEngine/2003-01-04-PhiTest.ll  Sun Dec 31 
00:01:59 2006
@@ -1,4 +1,4 @@
-; RUN: llvm-as -f %s -o %t.bc
+; RUN: llvm-upgrade < %s | llvm-as -f -o %t.bc
 ; RUN: lli %t.bc > /dev/null
 
 int %main() {


Index: llvm/test/Regression/ExecutionEngine/2003-05-11-PHIRegAllocBug.ll
diff -u llvm/test/Regression/ExecutionEngine/2003-05-11-PHIRegAllocBug.ll:1.3 
llvm/test/Regression/ExecutionEngine/2003-05-11-PHIRegAllocBug.ll:1.4
--- llvm/test/Regression/ExecutionEngine/2003-05-11-PHIRegAllocBug.ll:1.3   
Sat Nov  6 17:32:43 2004
+++ llvm/test/Regression/ExecutionEngine/2003-05-11-PHIRegAllocBug.ll   Sun Dec 
31 00:01:59 2006
@@ -1,4 +1,4 @@
-; RUN: llvm-as -f %s -o %t.bc
+; RUN: llvm-upgrade < %s | llvm-as -f -o %t.bc
 ; RUN: lli %t.bc > /dev/null
 
 target endian = little


Index: llvm/test/Regression/ExecutionEngine/2003-06-05-PHIBug.ll
diff -u llvm/test/Regression/ExecutionEngine/2003-06-05-PHIBug.ll:1.2 
llvm/test/Regression/ExecutionEngine/2003-06-05-PHIBug.ll:1.3
--- llvm/test/Regression/ExecutionEngine/2003-06-05-PHIBug.ll:1.2   Sat Nov 
 6 17:32:43 2004
+++ llvm/test/Regression/ExecutionEngine/2003-06-05-PHIBug.ll   Sun Dec 31 
00:01:59 2006
@@ -1,4 +1,4 @@
-; RUN: llvm-as -f %s -o %t.bc
+; RUN: llvm-upgrade < %s | llvm-as -f -o %t.bc
 ; RUN: lli %t.bc > /dev/null
 
 ; Testcase distilled from 256.bzip2.


Index: llvm/test/Regression/ExecutionEngine/2003-08-15-AllocaAssertion.ll
diff -u llvm/test/Regression/ExecutionEngine/2003-08-15-AllocaAssertion.ll:1.2 
llvm/test/Regression/ExecutionEngine/2003-08-15-AllocaAssertion.ll:1.3
--- llvm/test/Regression/ExecutionEngine/2003-08-15-AllocaAssertion.ll:1.2  
Sat Nov  6 17:32:43 2004
+++ llvm/test/Regression/ExecutionEngine/2003-08-15-AllocaAssertion.ll  Sun Dec 
31 00:01:59 2006
@@ -1,4 +1,4 @@
-; RUN: llvm-as -f %s -o %t.bc
+; RUN: llvm-upgrade < %s | llvm-as -f -o %t.bc
 ; RUN: lli %t.bc > /dev/null
 
 ; This testcase failed to work because two variable sized allocas confused the


Index: 
llvm/test/Regression/ExecutionEngine/2003-08-23-RegisterAllocatePhysReg.ll
diff -u 
llvm/test/Regression/ExecutionEngine/2003-08-23-RegisterAllocatePhysReg.ll:1.2 
llvm/test/Regression/ExecutionEngine/2003-08-23-RegisterAllocatePhysReg.ll:1.3
--- 
llvm/test/Regression/ExecutionEngine/2003-08-23-RegisterAllocatePhysReg.ll:1.2  
Sat Nov  6 17:32:43 2004
+++ llvm/test/Regression/ExecutionEngine/2003-08-23-RegisterAll

[llvm-commits] CVS: llvm/test/Regression/Assembler/2002-07-08-HugePerformanceProblem.llx 2003-04-15-ConstantInitAssertion.llx 2003-05-21-MalformedShiftCrash.llx 2003-05-21-MalformedStructCrash.llx 200

[Reid Spencer]

Changes in directory llvm/test/Regression/Assembler:

2002-07-08-HugePerformanceProblem.llx updated: 1.4 -> 1.5
2003-04-15-ConstantInitAssertion.llx updated: 1.2 -> 1.3
2003-05-21-MalformedShiftCrash.llx updated: 1.3 -> 1.4
2003-05-21-MalformedStructCrash.llx updated: 1.3 -> 1.4
2003-11-12-ConstantExprCast.llx updated: 1.3 -> 1.4
2003-12-30-TypeMapInvalidMemory.llx updated: 1.3 -> 1.4
2004-10-22-BCWriterUndefBug.llx updated: 1.2 -> 1.3
---
Log message:

For PR950: http://llvm.org/PR950 :
Update the test suite to accommodate the change from signed integer types
to signless integer types. The changes were of only a few kinds:

1. Make sure llvm-upgrade is run on the source which does the bulk of the
   changes automatically.

2. Change things like "grep 'int'" to "grep 'i32'"

3. In several tests bitcasting caused the same name to be reused in the
   same type plane. These had to be manually fixed. The fix was (generally)
   to leave the bitcast and provide the instruction with a new name. This
   should not affect the semantics of the test. In a few cases, the 
   bitcasts were known to be superfluous and irrelevant to the test case
   so they were removed.

4. One test case uses a bytecode file which needed to be updated to the
   latest bytecode format.



---
Diffs of the changes:  (+10 -9)

 2002-07-08-HugePerformanceProblem.llx |2 +-
 2003-04-15-ConstantInitAssertion.llx  |2 +-
 2003-05-21-MalformedShiftCrash.llx|2 +-
 2003-05-21-MalformedStructCrash.llx   |2 +-
 2003-11-12-ConstantExprCast.llx   |6 +++---
 2003-12-30-TypeMapInvalidMemory.llx   |3 ++-
 2004-10-22-BCWriterUndefBug.llx   |2 +-
 7 files changed, 10 insertions(+), 9 deletions(-)


Index: llvm/test/Regression/Assembler/2002-07-08-HugePerformanceProblem.llx
diff -u 
llvm/test/Regression/Assembler/2002-07-08-HugePerformanceProblem.llx:1.4 
llvm/test/Regression/Assembler/2002-07-08-HugePerformanceProblem.llx:1.5
--- llvm/test/Regression/Assembler/2002-07-08-HugePerformanceProblem.llx:1.4
Sat Nov  6 14:38:27 2004
+++ llvm/test/Regression/Assembler/2002-07-08-HugePerformanceProblem.llx
Sun Dec 31 00:01:59 2006
@@ -1,6 +1,6 @@
 ; This file takes about 48 __MINUTES__ to assemble using as.  This is WAY too
 ; long.  The type resolution code needs to be sped up a lot.   
-; RUN: ulimit -t 20; llvm-as < %s
+; RUN: ulimit -t 20; llvm-upgrade < %s | llvm-as
 
%ALL_INTERSECTIONS_METHOD = type int (%OBJECT*, %RAY*, %ISTACK*)*
%BBOX = type { %BBOX_VECT, %BBOX_VECT }


Index: llvm/test/Regression/Assembler/2003-04-15-ConstantInitAssertion.llx
diff -u llvm/test/Regression/Assembler/2003-04-15-ConstantInitAssertion.llx:1.2 
llvm/test/Regression/Assembler/2003-04-15-ConstantInitAssertion.llx:1.3
--- llvm/test/Regression/Assembler/2003-04-15-ConstantInitAssertion.llx:1.2 
Thu Oct 23 10:47:23 2003
+++ llvm/test/Regression/Assembler/2003-04-15-ConstantInitAssertion.llx Sun Dec 
31 00:01:59 2006
@@ -1,4 +1,4 @@
-; RUN: (llvm-as < %s 2>&1) | grep Expected
+; RUN: llvm-upgrade < %s | llvm-as 2>&1 >/dev/null | grep Expected
 ; Test the case of a misformed constant initializer
 ; This should cause an assembler error, not an assertion failure!
 %X = constant {int} { float 1.0 }


Index: llvm/test/Regression/Assembler/2003-05-21-MalformedShiftCrash.llx
diff -u llvm/test/Regression/Assembler/2003-05-21-MalformedShiftCrash.llx:1.3 
llvm/test/Regression/Assembler/2003-05-21-MalformedShiftCrash.llx:1.4
--- llvm/test/Regression/Assembler/2003-05-21-MalformedShiftCrash.llx:1.3   
Sat Dec  2 14:34:08 2006
+++ llvm/test/Regression/Assembler/2003-05-21-MalformedShiftCrash.llx   Sun Dec 
31 00:01:59 2006
@@ -1,4 +1,4 @@
 ; Found by inspection of the code
-; RUN: llvm-upgrade < %s | llvm-as -o /dev/null -f 2>&1 | grep "Shift constant 
expression"
+; RUN: llvm-upgrade < %s | llvm-as 2>&1 > /dev/null | grep "Shift constant 
expression"
 
 global int shr (float 1.0, ubyte 2)


Index: llvm/test/Regression/Assembler/2003-05-21-MalformedStructCrash.llx
diff -u llvm/test/Regression/Assembler/2003-05-21-MalformedStructCrash.llx:1.3 
llvm/test/Regression/Assembler/2003-05-21-MalformedStructCrash.llx:1.4
--- llvm/test/Regression/Assembler/2003-05-21-MalformedStructCrash.llx:1.3  
Mon Sep 15 15:01:21 2003
+++ llvm/test/Regression/Assembler/2003-05-21-MalformedStructCrash.llx  Sun Dec 
31 00:01:59 2006
@@ -1,4 +1,4 @@
 ; Found by inspection of the code
-; RUN: llvm-as < %s 2>&1 | grep "Illegal"
+; RUN: llvm-upgrade < %s | llvm-as 2>&1 > /dev/null | grep "Illegal"
 
 global {} { int 7, float 1.0, int 7, int 8 }


Index: llvm/test/Regression/Assembler/2003-11-12-ConstantExprCast.llx
diff -u llvm/test/Regression/Assembler/2003-11-12-ConstantExprCast.llx:1.3 
llvm/test/Regression/Assembler/2003-11-12-ConstantExprCast.llx:1.4
--- llvm/test/Regression/Assembler/2003-11-12-ConstantExprCast.llx:1.3  Fri Dec 
 1 22:23:08 2006
+++ llvm/test/Regression/Assembler/2003-11-12-ConstantExprCast.llx  Sun Dec 
31 00:01:59 20

[llvm-commits] CVS: llvm/test/Regression/C++Frontend/2006-09-27-Debug-Protection.cpp

[Reid Spencer]

Changes in directory llvm/test/Regression/C++Frontend:

2006-09-27-Debug-Protection.cpp updated: 1.2 -> 1.3
---
Log message:

For PR950: http://llvm.org/PR950 :
Update the test suite to accommodate the change from signed integer types
to signless integer types. The changes were of only a few kinds:

1. Make sure llvm-upgrade is run on the source which does the bulk of the
   changes automatically.

2. Change things like "grep 'int'" to "grep 'i32'"

3. In several tests bitcasting caused the same name to be reused in the
   same type plane. These had to be manually fixed. The fix was (generally)
   to leave the bitcast and provide the instruction with a new name. This
   should not affect the semantics of the test. In a few cases, the 
   bitcasts were known to be superfluous and irrelevant to the test case
   so they were removed.

4. One test case uses a bytecode file which needed to be updated to the
   latest bytecode format.



---
Diffs of the changes:  (+2 -2)

 2006-09-27-Debug-Protection.cpp |4 ++--
 1 files changed, 2 insertions(+), 2 deletions(-)


Index: llvm/test/Regression/C++Frontend/2006-09-27-Debug-Protection.cpp
diff -u llvm/test/Regression/C++Frontend/2006-09-27-Debug-Protection.cpp:1.2 
llvm/test/Regression/C++Frontend/2006-09-27-Debug-Protection.cpp:1.3
--- llvm/test/Regression/C++Frontend/2006-09-27-Debug-Protection.cpp:1.2
Wed Nov  8 14:16:05 2006
+++ llvm/test/Regression/C++Frontend/2006-09-27-Debug-Protection.cppSun Dec 
31 00:01:59 2006
@@ -1,6 +1,6 @@
 // XFAIL: llvmgcc3
-// RUN: %llvmgxx -O0 -emit-llvm -S -g -o - %s | grep 'uint 1,' &&
-// RUN: %llvmgxx -O0 -emit-llvm -S -g -o - %s | grep 'uint 2,'
+// RUN: %llvmgxx -O0 -emit-llvm -S -g -o - %s | grep 'i32 1,' &&
+// RUN: %llvmgxx -O0 -emit-llvm -S -g -o - %s | grep 'i32 2,'
 
 class A {
 public:



___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

[llvm-commits] CVS: llvm/test/Regression/Transforms/ConstProp/2006-11-30-vector-cast.ll 2006-12-01-bool-casts.ll bitcast.ll float-to-ptr-cast.ll

[Reid Spencer]

Changes in directory llvm/test/Regression/Transforms/ConstProp:

2006-11-30-vector-cast.ll updated: 1.3 -> 1.4
2006-12-01-bool-casts.ll updated: 1.2 -> 1.3
bitcast.ll updated: 1.1 -> 1.2
float-to-ptr-cast.ll updated: 1.2 -> 1.3
---
Log message:

For PR950: http://llvm.org/PR950 :
Update the test suite to accommodate the change from signed integer types
to signless integer types. The changes were of only a few kinds:

1. Make sure llvm-upgrade is run on the source which does the bulk of the
   changes automatically.

2. Change things like "grep 'int'" to "grep 'i32'"

3. In several tests bitcasting caused the same name to be reused in the
   same type plane. These had to be manually fixed. The fix was (generally)
   to leave the bitcast and provide the instruction with a new name. This
   should not affect the semantics of the test. In a few cases, the 
   bitcasts were known to be superfluous and irrelevant to the test case
   so they were removed.

4. One test case uses a bytecode file which needed to be updated to the
   latest bytecode format.



---
Diffs of the changes:  (+6 -6)

 2006-11-30-vector-cast.ll |2 +-
 2006-12-01-bool-casts.ll  |4 ++--
 bitcast.ll|4 ++--
 float-to-ptr-cast.ll  |2 +-
 4 files changed, 6 insertions(+), 6 deletions(-)


Index: llvm/test/Regression/Transforms/ConstProp/2006-11-30-vector-cast.ll
diff -u llvm/test/Regression/Transforms/ConstProp/2006-11-30-vector-cast.ll:1.3 
llvm/test/Regression/Transforms/ConstProp/2006-11-30-vector-cast.ll:1.4
--- llvm/test/Regression/Transforms/ConstProp/2006-11-30-vector-cast.ll:1.3 
Tue Dec 19 16:42:17 2006
+++ llvm/test/Regression/Transforms/ConstProp/2006-11-30-vector-cast.ll Sun Dec 
31 00:01:59 2006
@@ -1,4 +1,4 @@
-; RUN: llvm-upgrade < %s | llvm-as | opt -constprop | llvm-dis | grep 'uint 
-1' &&
+; RUN: llvm-upgrade < %s | llvm-as | opt -constprop | llvm-dis | grep 'i32 -1' 
&&
 ; RUN: llvm-upgrade < %s | llvm-as | opt -constprop | llvm-dis | not grep 
zeroinitializer
 
 < 4 x uint> %test() {


Index: llvm/test/Regression/Transforms/ConstProp/2006-12-01-bool-casts.ll
diff -u llvm/test/Regression/Transforms/ConstProp/2006-12-01-bool-casts.ll:1.2 
llvm/test/Regression/Transforms/ConstProp/2006-12-01-bool-casts.ll:1.3
--- llvm/test/Regression/Transforms/ConstProp/2006-12-01-bool-casts.ll:1.2  
Fri Dec  1 22:23:09 2006
+++ llvm/test/Regression/Transforms/ConstProp/2006-12-01-bool-casts.ll  Sun Dec 
31 00:01:59 2006
@@ -1,5 +1,5 @@
-; RUN: llvm-upgrade < %s | llvm-as | opt -constprop | llvm-dis | grep 'ret int 
-1' &&
-; RUN: llvm-upgrade < %s | llvm-as | opt -constprop | llvm-dis | grep 'ret 
uint 1'
+; RUN: llvm-upgrade < %s | llvm-as | opt -constprop | llvm-dis | grep 'ret i32 
-1' &&
+; RUN: llvm-upgrade < %s | llvm-as | opt -constprop | llvm-dis | grep 'ret i32 
1'
 
 int %test1() {
   %A = sext bool true to int


Index: llvm/test/Regression/Transforms/ConstProp/bitcast.ll
diff -u llvm/test/Regression/Transforms/ConstProp/bitcast.ll:1.1 
llvm/test/Regression/Transforms/ConstProp/bitcast.ll:1.2
--- llvm/test/Regression/Transforms/ConstProp/bitcast.ll:1.1Mon Dec 11 
12:29:07 2006
+++ llvm/test/Regression/Transforms/ConstProp/bitcast.llSun Dec 31 
00:01:59 2006
@@ -1,4 +1,4 @@
-; RUN: llvm-as < %s | llvm-dis &&
-; RUN: llvm-as < %s | llvm-dis | grep 0x36A0
+; RUN: llvm-upgrade < %s | llvm-as | llvm-dis &&
+; RUN: llvm-upgrade < %s | llvm-as | llvm-dis | grep 0x36A0
 
 %A = global float bitcast (int 1 to float)


Index: llvm/test/Regression/Transforms/ConstProp/float-to-ptr-cast.ll
diff -u llvm/test/Regression/Transforms/ConstProp/float-to-ptr-cast.ll:1.2 
llvm/test/Regression/Transforms/ConstProp/float-to-ptr-cast.ll:1.3
--- llvm/test/Regression/Transforms/ConstProp/float-to-ptr-cast.ll:1.2  Fri Dec 
 1 22:23:09 2006
+++ llvm/test/Regression/Transforms/ConstProp/float-to-ptr-cast.ll  Sun Dec 
31 00:01:59 2006
@@ -1,4 +1,4 @@
-; RUN: llvm-upgrade < %s | llvm-as | opt -constprop | llvm-dis | grep -F 'ret 
int* null' | wc -l | grep 2
+; RUN: llvm-upgrade < %s | llvm-as | opt -constprop | llvm-dis | grep -F 'ret 
i32* null' | wc -l | grep 2
 int* %test1() {
   %X = cast float 0.0 to int*
   ret int* %X



___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

[llvm-commits] CVS: llvm/test/Regression/CodeGen/Alpha/zapnot.ll

[Reid Spencer]

Changes in directory llvm/test/Regression/CodeGen/Alpha:

zapnot.ll updated: 1.2 -> 1.3
---
Log message:

For PR950: http://llvm.org/PR950 :
Update the test suite to accommodate the change from signed integer types
to signless integer types. The changes were of only a few kinds:

1. Make sure llvm-upgrade is run on the source which does the bulk of the
   changes automatically.

2. Change things like "grep 'int'" to "grep 'i32'"

3. In several tests bitcasting caused the same name to be reused in the
   same type plane. These had to be manually fixed. The fix was (generally)
   to leave the bitcast and provide the instruction with a new name. This
   should not affect the semantics of the test. In a few cases, the 
   bitcasts were known to be superfluous and irrelevant to the test case
   so they were removed.

4. One test case uses a bytecode file which needed to be updated to the
   latest bytecode format.



---
Diffs of the changes:  (+4 -5)

 zapnot.ll |9 -
 1 files changed, 4 insertions(+), 5 deletions(-)


Index: llvm/test/Regression/CodeGen/Alpha/zapnot.ll
diff -u llvm/test/Regression/CodeGen/Alpha/zapnot.ll:1.2 
llvm/test/Regression/CodeGen/Alpha/zapnot.ll:1.3
--- llvm/test/Regression/CodeGen/Alpha/zapnot.ll:1.2Fri Dec  1 22:23:08 2006
+++ llvm/test/Regression/CodeGen/Alpha/zapnot.llSun Dec 31 00:01:59 2006
@@ -1,11 +1,10 @@
 ; Make sure this testcase codegens to the bic instruction
-; RUN: llvm-upgrade < %s | llvm-as | llc -march=alpha | grep 'zapnot'
+; RUN: llvm-as < %s | llc -march=alpha | grep 'zapnot'
 
 implementation   ; Functions:
 
-ushort %foo(long %y) {
+define i16 @zext %foo(i64 %y) {
 entry:
-%tmp.1 = cast long %y to ushort ;  [#uses=1]
-ret ushort %tmp.1
+%tmp.1 = trunc i64 %y to i16 ;  [#uses=1]
+ret i16 %tmp.1
 }
-



___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

[llvm-commits] CVS: llvm/test/Regression/Transforms/DeadStoreElimination/2004-12-28-PartialStore.ll 2006-06-27-AST-Remove.ll

[Reid Spencer]

Changes in directory llvm/test/Regression/Transforms/DeadStoreElimination:

2004-12-28-PartialStore.ll updated: 1.3 -> 1.4
2006-06-27-AST-Remove.ll updated: 1.2 -> 1.3
---
Log message:

For PR950: http://llvm.org/PR950 :
Update the test suite to accommodate the change from signed integer types
to signless integer types. The changes were of only a few kinds:

1. Make sure llvm-upgrade is run on the source which does the bulk of the
   changes automatically.

2. Change things like "grep 'int'" to "grep 'i32'"

3. In several tests bitcasting caused the same name to be reused in the
   same type plane. These had to be manually fixed. The fix was (generally)
   to leave the bitcast and provide the instruction with a new name. This
   should not affect the semantics of the test. In a few cases, the 
   bitcasts were known to be superfluous and irrelevant to the test case
   so they were removed.

4. One test case uses a bytecode file which needed to be updated to the
   latest bytecode format.



---
Diffs of the changes:  (+13 -13)

 2004-12-28-PartialStore.ll |2 +-
 2006-06-27-AST-Remove.ll   |   24 
 2 files changed, 13 insertions(+), 13 deletions(-)


Index: 
llvm/test/Regression/Transforms/DeadStoreElimination/2004-12-28-PartialStore.ll
diff -u 
llvm/test/Regression/Transforms/DeadStoreElimination/2004-12-28-PartialStore.ll:1.3
 
llvm/test/Regression/Transforms/DeadStoreElimination/2004-12-28-PartialStore.ll:1.4
--- 
llvm/test/Regression/Transforms/DeadStoreElimination/2004-12-28-PartialStore.ll:1.3
 Fri Dec  1 22:23:09 2006
+++ 
llvm/test/Regression/Transforms/DeadStoreElimination/2004-12-28-PartialStore.ll 
Sun Dec 31 00:01:59 2006
@@ -1,4 +1,4 @@
-; RUN: llvm-upgrade < %s | llvm-as | opt -dse | llvm-dis | grep 'store int 
1234567'
+; RUN: llvm-upgrade < %s | llvm-as | opt -dse | llvm-dis | grep 'store i32 
1234567'
 
 ; Do not delete stores that are only partially killed.
 


Index: 
llvm/test/Regression/Transforms/DeadStoreElimination/2006-06-27-AST-Remove.ll
diff -u 
llvm/test/Regression/Transforms/DeadStoreElimination/2006-06-27-AST-Remove.ll:1.2
 
llvm/test/Regression/Transforms/DeadStoreElimination/2006-06-27-AST-Remove.ll:1.3
--- 
llvm/test/Regression/Transforms/DeadStoreElimination/2006-06-27-AST-Remove.ll:1.2
   Fri Dec  1 22:23:09 2006
+++ 
llvm/test/Regression/Transforms/DeadStoreElimination/2006-06-27-AST-Remove.ll   
Sun Dec 31 00:01:59 2006
@@ -755,22 +755,22 @@
%tmp.i = load %struct.TType** %DP_TTable; 
<%struct.TType*> [#uses=1]
%tmp.i7.b = load bool* %TTSize.b;  [#uses=1]
%tmp1.i = select bool %tmp.i7.b, uint 6000, uint 0  ; 
 [#uses=1]
-   %tmp.i = getelementptr %struct.TType* %tmp.i, int 0, uint 0 
;  [#uses=1]
-   call void %llvm.memset.i32( sbyte* %tmp.i, ubyte 0, uint %tmp1.i, uint 
4 )
+   %tmp.i.sb = getelementptr %struct.TType* %tmp.i, int 0, uint 0  
;  [#uses=1]
+   call void %llvm.memset.i32( sbyte* %tmp.i.sb, ubyte 0, uint %tmp1.i, 
uint 4 )
%tmp2.i = load %struct.TType** %AS_TTable   ; 
<%struct.TType*> [#uses=1]
%tmp3.i8.b = load bool* %TTSize.b   ;  [#uses=1]
%tmp4.i = select bool %tmp3.i8.b, uint 6000, uint 0 ; 
 [#uses=1]
%tmp2.i = getelementptr %struct.TType* %tmp2.i, int 0, uint 0   
;  [#uses=1]
call void %llvm.memset.i32( sbyte* %tmp2.i, ubyte 0, uint %tmp4.i, uint 
4 )
-   %tmp.i = load %struct.QTType** %QS_TTable   ; 
<%struct.QTType*> [#uses=1]
+   %tmp.i.QTT = load %struct.QTType** %QS_TTable   ; 
<%struct.QTType*> [#uses=1]
%tmp5.i9.b = load bool* %TTSize.b   ;  [#uses=1]
%tmp6.i10 = select bool %tmp5.i9.b, uint 4800, uint 0   
;  [#uses=1]
-   %tmp7.i = getelementptr %struct.QTType* %tmp.i, int 0, uint 0   
;  [#uses=1]
+   %tmp7.i = getelementptr %struct.QTType* %tmp.i.QTT, int 0, uint 0   
;  [#uses=1]
call void %llvm.memset.i32( sbyte* %tmp7.i, ubyte 0, uint %tmp6.i10, 
uint 4 )
-   %tmp.i = load %struct.ECacheType** %ECache  ; 
<%struct.ECacheType*> [#uses=1]
+   %tmp.i.ECache = load %struct.ECacheType** %ECache   ; 
<%struct.ECacheType*> [#uses=1]
%tmp.i14.b = load bool* %ECacheSize.b   ;  [#uses=1]
%tmp1.i16 = select bool %tmp.i14.b, uint 1200, uint 0   
;  [#uses=1]
-   %tmp.i17 = cast %struct.ECacheType* %tmp.i to sbyte*; 
 [#uses=1]
+   %tmp.i17 = cast %struct.ECacheType* %tmp.i.ECache to sbyte* 
;  [#uses=1]
call void %llvm.memset.i32( sbyte* %tmp.i17, ubyte 0, uint %tmp1.i16, 
uint 4 )
call void %llvm.memset.i32( sbyte* cast ([300 x int]* %rootlosers to 
sbyte*), ubyte 0, uint 1200, uint 4 )
%tmp234.b = load bool* %is_pondering.b  ;  [#uses=1]
@@ -800,12 +800,12 @@
%tmp1.b.i = load 

[llvm-commits] CVS: llvm/test/Regression/Transforms/ScalarRepl/2006-11-07-InvalidArrayPromote.ll union-fp-int.ll union-pointer.ll

[Reid Spencer]

Changes in directory llvm/test/Regression/Transforms/ScalarRepl:

2006-11-07-InvalidArrayPromote.ll updated: 1.2 -> 1.3
union-fp-int.ll updated: 1.3 -> 1.4
union-pointer.ll updated: 1.3 -> 1.4
---
Log message:

For PR950: http://llvm.org/PR950 :
Update the test suite to accommodate the change from signed integer types
to signless integer types. The changes were of only a few kinds:

1. Make sure llvm-upgrade is run on the source which does the bulk of the
   changes automatically.

2. Change things like "grep 'int'" to "grep 'i32'"

3. In several tests bitcasting caused the same name to be reused in the
   same type plane. These had to be manually fixed. The fix was (generally)
   to leave the bitcast and provide the instruction with a new name. This
   should not affect the semantics of the test. In a few cases, the 
   bitcasts were known to be superfluous and irrelevant to the test case
   so they were removed.

4. One test case uses a bytecode file which needed to be updated to the
   latest bytecode format.



---
Diffs of the changes:  (+3 -3)

 2006-11-07-InvalidArrayPromote.ll |2 +-
 union-fp-int.ll   |2 +-
 union-pointer.ll  |2 +-
 3 files changed, 3 insertions(+), 3 deletions(-)


Index: 
llvm/test/Regression/Transforms/ScalarRepl/2006-11-07-InvalidArrayPromote.ll
diff -u 
llvm/test/Regression/Transforms/ScalarRepl/2006-11-07-InvalidArrayPromote.ll:1.2
 
llvm/test/Regression/Transforms/ScalarRepl/2006-11-07-InvalidArrayPromote.ll:1.3
--- 
llvm/test/Regression/Transforms/ScalarRepl/2006-11-07-InvalidArrayPromote.ll:1.2
Fri Dec  1 22:23:10 2006
+++ 
llvm/test/Regression/Transforms/ScalarRepl/2006-11-07-InvalidArrayPromote.ll
Sun Dec 31 00:02:00 2006
@@ -1,5 +1,5 @@
 ; RUN: llvm-upgrade < %s | llvm-as | opt -scalarrepl | llvm-dis &&
-; RUN: llvm-upgrade < %s | llvm-as | opt -scalarrepl | llvm-dis | grep -F 
'alloca [2 x <4 x int>]'
+; RUN: llvm-upgrade < %s | llvm-as | opt -scalarrepl | llvm-dis | grep -F 
'alloca [2 x <4 x i32>]'
 
 int %func(<4 x float> %v0, <4 x float> %v1) {
%vsiidx = alloca [2 x <4 x int>], align 16  ; <[2 x <4 x 
int>]*> [#uses=3]


Index: llvm/test/Regression/Transforms/ScalarRepl/union-fp-int.ll
diff -u llvm/test/Regression/Transforms/ScalarRepl/union-fp-int.ll:1.3 
llvm/test/Regression/Transforms/ScalarRepl/union-fp-int.ll:1.4
--- llvm/test/Regression/Transforms/ScalarRepl/union-fp-int.ll:1.3  Fri Dec 
15 01:32:49 2006
+++ llvm/test/Regression/Transforms/ScalarRepl/union-fp-int.ll  Sun Dec 31 
00:02:00 2006
@@ -1,5 +1,5 @@
 ; RUN: llvm-upgrade < %s | llvm-as | opt -scalarrepl | llvm-dis | not grep 
alloca &&
-; RUN: llvm-upgrade < %s | llvm-as | opt -scalarrepl | llvm-dis | grep 
'bitcast.*float.*int'
+; RUN: llvm-upgrade < %s | llvm-as | opt -scalarrepl | llvm-dis | grep 
'bitcast.*float.*i32'
 
 int %test(float %X) {
 %X_addr = alloca float


Index: llvm/test/Regression/Transforms/ScalarRepl/union-pointer.ll
diff -u llvm/test/Regression/Transforms/ScalarRepl/union-pointer.ll:1.3 
llvm/test/Regression/Transforms/ScalarRepl/union-pointer.ll:1.4
--- llvm/test/Regression/Transforms/ScalarRepl/union-pointer.ll:1.3 Fri Dec 
 1 22:23:10 2006
+++ llvm/test/Regression/Transforms/ScalarRepl/union-pointer.ll Sun Dec 31 
00:02:00 2006
@@ -1,5 +1,5 @@
 ; RUN: llvm-upgrade < %s | llvm-as | opt -scalarrepl | llvm-dis | not grep 
alloca &&
-; RUN: llvm-upgrade < %s | llvm-as | opt -scalarrepl | llvm-dis | grep 'ret 
sbyte'
+; RUN: llvm-upgrade < %s | llvm-as | opt -scalarrepl | llvm-dis | grep 'ret i8'
 
 ; PR892
 



___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

[llvm-commits] CVS: llvm/test/Regression/Transforms/IndVarsSimplify/2005-02-26-ExitValueCompute.ll 2006-03-31-NegativeStride.ll variable-stride-ivs.ll

[Reid Spencer]

Changes in directory llvm/test/Regression/Transforms/IndVarsSimplify:

2005-02-26-ExitValueCompute.ll updated: 1.2 -> 1.3
2006-03-31-NegativeStride.ll updated: 1.2 -> 1.3
variable-stride-ivs.ll updated: 1.2 -> 1.3
---
Log message:

For PR950: http://llvm.org/PR950 :
Update the test suite to accommodate the change from signed integer types
to signless integer types. The changes were of only a few kinds:

1. Make sure llvm-upgrade is run on the source which does the bulk of the
   changes automatically.

2. Change things like "grep 'int'" to "grep 'i32'"

3. In several tests bitcasting caused the same name to be reused in the
   same type plane. These had to be manually fixed. The fix was (generally)
   to leave the bitcast and provide the instruction with a new name. This
   should not affect the semantics of the test. In a few cases, the 
   bitcasts were known to be superfluous and irrelevant to the test case
   so they were removed.

4. One test case uses a bytecode file which needed to be updated to the
   latest bytecode format.



---
Diffs of the changes:  (+6 -6)

 2005-02-26-ExitValueCompute.ll |2 +-
 2006-03-31-NegativeStride.ll   |2 +-
 variable-stride-ivs.ll |8 
 3 files changed, 6 insertions(+), 6 deletions(-)


Index: 
llvm/test/Regression/Transforms/IndVarsSimplify/2005-02-26-ExitValueCompute.ll
diff -u 
llvm/test/Regression/Transforms/IndVarsSimplify/2005-02-26-ExitValueCompute.ll:1.2
 
llvm/test/Regression/Transforms/IndVarsSimplify/2005-02-26-ExitValueCompute.ll:1.3
--- 
llvm/test/Regression/Transforms/IndVarsSimplify/2005-02-26-ExitValueCompute.ll:1.2
  Fri Dec  1 22:23:09 2006
+++ 
llvm/test/Regression/Transforms/IndVarsSimplify/2005-02-26-ExitValueCompute.ll  
Sun Dec 31 00:01:59 2006
@@ -1,4 +1,4 @@
-; RUN: llvm-upgrade < %s | llvm-as | opt -indvars | llvm-dis | grep 'ret int 
152'
+; RUN: llvm-upgrade < %s | llvm-as | opt -indvars | llvm-dis | grep 'ret i32 
152'
 
 int %main() {
 entry:


Index: 
llvm/test/Regression/Transforms/IndVarsSimplify/2006-03-31-NegativeStride.ll
diff -u 
llvm/test/Regression/Transforms/IndVarsSimplify/2006-03-31-NegativeStride.ll:1.2
 
llvm/test/Regression/Transforms/IndVarsSimplify/2006-03-31-NegativeStride.ll:1.3
--- 
llvm/test/Regression/Transforms/IndVarsSimplify/2006-03-31-NegativeStride.ll:1.2
Fri Dec  1 22:23:09 2006
+++ 
llvm/test/Regression/Transforms/IndVarsSimplify/2006-03-31-NegativeStride.ll
Sun Dec 31 00:01:59 2006
@@ -1,4 +1,4 @@
-; RUN: llvm-upgrade < %s | llvm-as | opt -indvars | llvm-dis | grep 'ret int 
27'
+; RUN: llvm-upgrade < %s | llvm-as | opt -indvars | llvm-dis | grep 'ret i32 
27'
 
 ; Make sure to compute the right exit value based on negative strides.
 ; PR726


Index: llvm/test/Regression/Transforms/IndVarsSimplify/variable-stride-ivs.ll
diff -u 
llvm/test/Regression/Transforms/IndVarsSimplify/variable-stride-ivs.ll:1.2 
llvm/test/Regression/Transforms/IndVarsSimplify/variable-stride-ivs.ll:1.3
--- llvm/test/Regression/Transforms/IndVarsSimplify/variable-stride-ivs.ll:1.2  
Fri Dec  1 22:23:09 2006
+++ llvm/test/Regression/Transforms/IndVarsSimplify/variable-stride-ivs.ll  
Sun Dec 31 00:01:59 2006
@@ -1,4 +1,4 @@
-; RUN: llvm-upgrade < %s | llvm-as | opt -indvars -instcombine | llvm-dis | 
grep 'store int 0'
+; RUN: llvm-upgrade < %s | llvm-as | opt -indvars -instcombine | llvm-dis | 
grep 'store i32 0'
 ; Test that -indvars can reduce variable stride IVs.  If it can reduce variable
 ; stride iv's, it will make %iv. and %m.0.0 isomorphic to each other without 
 ; cycles, allowing the tmp.21 subtraction to be eliminated.
@@ -19,7 +19,7 @@
 br label %no_exit
 
 no_exit:; preds = %no_exit, %no_exit.preheader
-%iv. = phi uint [ 0, %no_exit.preheader ], [ %iv..inc, %no_exit ]  
 ;  [#uses=1]
+%iv.ui = phi uint [ 0, %no_exit.preheader ], [ %iv..inc.ui, %no_exit ] 
  ;  [#uses=1]
 %iv. = phi int [ %tmp.5, %no_exit.preheader ], [ %iv..inc, %no_exit ]  
 ;  [#uses=2]
 %m.0.0 = phi int [ %tmp.5, %no_exit.preheader ], [ %tmp.24, %no_exit ] 
 ;  [#uses=2]
 store int 2, int* %tmp.16
@@ -27,8 +27,8 @@
 store int %tmp.21, int* %data
 %tmp.24 = add int %m.0.0, %tmp.9;  [#uses=1]
 %iv..inc = add int %tmp.9, %iv. ;  [#uses=1]
-%iv..inc = add uint %iv., 1 ;  [#uses=2]
-%iv..inc1 = cast uint %iv..inc to int   ;  [#uses=1]
+%iv..inc.ui = add uint %iv.ui, 1 ;  [#uses=2]
+%iv..inc1 = cast uint %iv..inc.ui to int   ;  [#uses=1]
 %tmp.12 = setlt int %iv..inc1, %tmp.2   ;  [#uses=1]
 br bool %tmp.12, label %no_exit, label %return.loopexit
 



___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

[llvm-commits] CVS: llvm/test/Regression/Linker/2002-07-17-GlobalFail.ll 2002-07-17-LinkTest2.ll 2002-08-20-ConstantExpr.ll 2003-04-26-NullPtrLinkProblem.ll 2003-05-15-TypeProblem.ll 2003-06-02-TypeRe

[Reid Spencer]

Changes in directory llvm/test/Regression/Linker:

2002-07-17-GlobalFail.ll updated: 1.3 -> 1.4
2002-07-17-LinkTest2.ll updated: 1.4 -> 1.5
2002-08-20-ConstantExpr.ll updated: 1.4 -> 1.5
2003-04-26-NullPtrLinkProblem.ll updated: 1.4 -> 1.5
2003-05-15-TypeProblem.ll updated: 1.3 -> 1.4
2003-06-02-TypeResolveProblem.ll updated: 1.3 -> 1.4
2003-06-02-TypeResolveProblem2.ll updated: 1.3 -> 1.4
2003-08-20-OpaqueTypeResolve.ll updated: 1.3 -> 1.4
2003-08-23-GlobalVarLinking.ll updated: 1.4 -> 1.5
2003-08-23-RecursiveOpaqueTypeResolve.ll updated: 1.4 -> 1.5
2003-10-21-ConflictingTypesTolerance.ll updated: 1.1 -> 1.2
2003-10-27-LinkOncePromote.ll updated: 1.2 -> 1.3
2004-02-17-WeakStrongLinkage.ll updated: 1.1 -> 1.2
2005-02-12-ConstantGlobals-2.ll updated: 1.1 -> 1.2
2005-02-12-ConstantGlobals.ll updated: 1.2 -> 1.3
2005-12-06-AppendingZeroLengthArrays.ll updated: 1.1 -> 1.2
2006-01-19-ConstantPacked.ll updated: 1.1 -> 1.2
2006-06-15-GlobalVarAlignment.ll updated: 1.1 -> 1.2
AppendingLinkage2.ll updated: 1.3 -> 1.4
ConstantGlobals1.ll updated: 1.2 -> 1.3
ConstantGlobals2.ll updated: 1.2 -> 1.3
ConstantGlobals3.ll updated: 1.2 -> 1.3
LinkOnce.ll updated: 1.3 -> 1.4
weakextern.ll updated: 1.2 -> 1.3
---
Log message:

For PR950: http://llvm.org/PR950 :
Update the test suite to accommodate the change from signed integer types
to signless integer types. The changes were of only a few kinds:

1. Make sure llvm-upgrade is run on the source which does the bulk of the
   changes automatically.

2. Change things like "grep 'int'" to "grep 'i32'"

3. In several tests bitcasting caused the same name to be reused in the
   same type plane. These had to be manually fixed. The fix was (generally)
   to leave the bitcast and provide the instruction with a new name. This
   should not affect the semantics of the test. In a few cases, the 
   bitcasts were known to be superfluous and irrelevant to the test case
   so they were removed.

4. One test case uses a bytecode file which needed to be updated to the
   latest bytecode format.



---
Diffs of the changes:  (+47 -47)

 2002-07-17-GlobalFail.ll |2 +-
 2002-07-17-LinkTest2.ll  |2 +-
 2002-08-20-ConstantExpr.ll   |2 +-
 2003-04-26-NullPtrLinkProblem.ll |2 +-
 2003-05-15-TypeProblem.ll|4 ++--
 2003-06-02-TypeResolveProblem.ll |4 ++--
 2003-06-02-TypeResolveProblem2.ll|4 ++--
 2003-08-20-OpaqueTypeResolve.ll  |4 ++--
 2003-08-23-GlobalVarLinking.ll   |4 ++--
 2003-08-23-RecursiveOpaqueTypeResolve.ll |4 ++--
 2003-10-21-ConflictingTypesTolerance.ll  |4 ++--
 2003-10-27-LinkOncePromote.ll|4 ++--
 2004-02-17-WeakStrongLinkage.ll  |4 ++--
 2005-02-12-ConstantGlobals-2.ll  |6 +++---
 2005-02-12-ConstantGlobals.ll|6 +++---
 2005-12-06-AppendingZeroLengthArrays.ll  |4 ++--
 2006-01-19-ConstantPacked.ll |2 +-
 2006-06-15-GlobalVarAlignment.ll |4 ++--
 AppendingLinkage2.ll |4 ++--
 ConstantGlobals1.ll  |4 ++--
 ConstantGlobals2.ll  |4 ++--
 ConstantGlobals3.ll  |2 +-
 LinkOnce.ll  |4 ++--
 weakextern.ll|   10 +-
 24 files changed, 47 insertions(+), 47 deletions(-)


Index: llvm/test/Regression/Linker/2002-07-17-GlobalFail.ll
diff -u llvm/test/Regression/Linker/2002-07-17-GlobalFail.ll:1.3 
llvm/test/Regression/Linker/2002-07-17-GlobalFail.ll:1.4
--- llvm/test/Regression/Linker/2002-07-17-GlobalFail.ll:1.3Mon Sep 15 
15:04:28 2003
+++ llvm/test/Regression/Linker/2002-07-17-GlobalFail.llSun Dec 31 
00:01:59 2006
@@ -1,4 +1,4 @@
-; RUN: llvm-as < %s > %t.bc
+; RUN: llvm-upgrade < %s | llvm-as > %t.bc
 ; RUN: echo | llvm-as > %t.tmp.bc
 ; RUN: llvm-link %t.tmp.bc %t.bc
 


Index: llvm/test/Regression/Linker/2002-07-17-LinkTest2.ll
diff -u llvm/test/Regression/Linker/2002-07-17-LinkTest2.ll:1.4 
llvm/test/Regression/Linker/2002-07-17-LinkTest2.ll:1.5
--- llvm/test/Regression/Linker/2002-07-17-LinkTest2.ll:1.4 Mon Sep 15 
15:04:28 2003
+++ llvm/test/Regression/Linker/2002-07-17-LinkTest2.ll Sun Dec 31 00:01:59 2006
@@ -1,7 +1,7 @@
 ; This fails linking when it is linked with an empty file as the first object 
file
 
 ; RUN: llvm-as > %t1.bc < /dev/null
-; RUN: llvm-as < %s > %t2.bc
+; RUN: llvm-upgrade < %s | llvm-as > %t2.bc
 ; RUN: llvm-link %t[12].bc
 
 %work = global int (int, int)* %zip


Index: llvm/test/Regression/Linker/2002-08-20-ConstantExpr.ll
diff -u llvm/test/Regression/Linker/2002-08-20-ConstantExpr.ll:1.4 
llvm/test/Regression/Linker/2002-08-20-ConstantExpr.ll:1.5
--- llvm/test/Regression/Linker/2002-08-20-ConstantExpr.ll:1.4  Mon Sep 15 
15:04:28 2003
+++ llvm/test/Regression/Linker/2002-08-20-ConstantExpr.ll  Sun Dec 31 
00:01:59 2006
@@ -1,7 +1,7 @@
 ; This fails lin

[llvm-commits] CVS: llvm/test/Regression/Transforms/LoopSimplify/2006-08-11-LoopSimplifyLongTime.ll.bc

[Reid Spencer]

Changes in directory llvm/test/Regression/Transforms/LoopSimplify:

2006-08-11-LoopSimplifyLongTime.ll.bc updated: 1.3 -> 1.4
---
Log message:

For PR950: http://llvm.org/PR950 :
Update the test suite to accommodate the change from signed integer types
to signless integer types. The changes were of only a few kinds:

1. Make sure llvm-upgrade is run on the source which does the bulk of the
   changes automatically.

2. Change things like "grep 'int'" to "grep 'i32'"

3. In several tests bitcasting caused the same name to be reused in the
   same type plane. These had to be manually fixed. The fix was (generally)
   to leave the bitcast and provide the instruction with a new name. This
   should not affect the semantics of the test. In a few cases, the 
   bitcasts were known to be superfluous and irrelevant to the test case
   so they were removed.

4. One test case uses a bytecode file which needed to be updated to the
   latest bytecode format.



---
Diffs of the changes:  (+0 -0)

 2006-08-11-LoopSimplifyLongTime.ll.bc |0 
 1 files changed


Index: 
llvm/test/Regression/Transforms/LoopSimplify/2006-08-11-LoopSimplifyLongTime.ll.bc



___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

[llvm-commits] CVS: llvm/test/Regression/Transforms/LoopUnswitch/2006-02-14-LoopSimplifyCrash.ll

[Reid Spencer]

Changes in directory llvm/test/Regression/Transforms/LoopUnswitch:

2006-02-14-LoopSimplifyCrash.ll updated: 1.2 -> 1.3
---
Log message:

For PR950: http://llvm.org/PR950 :
Update the test suite to accommodate the change from signed integer types
to signless integer types. The changes were of only a few kinds:

1. Make sure llvm-upgrade is run on the source which does the bulk of the
   changes automatically.

2. Change things like "grep 'int'" to "grep 'i32'"

3. In several tests bitcasting caused the same name to be reused in the
   same type plane. These had to be manually fixed. The fix was (generally)
   to leave the bitcast and provide the instruction with a new name. This
   should not affect the semantics of the test. In a few cases, the 
   bitcasts were known to be superfluous and irrelevant to the test case
   so they were removed.

4. One test case uses a bytecode file which needed to be updated to the
   latest bytecode format.



---
Diffs of the changes:  (+3 -3)

 2006-02-14-LoopSimplifyCrash.ll |6 +++---
 1 files changed, 3 insertions(+), 3 deletions(-)


Index: 
llvm/test/Regression/Transforms/LoopUnswitch/2006-02-14-LoopSimplifyCrash.ll
diff -u 
llvm/test/Regression/Transforms/LoopUnswitch/2006-02-14-LoopSimplifyCrash.ll:1.2
 
llvm/test/Regression/Transforms/LoopUnswitch/2006-02-14-LoopSimplifyCrash.ll:1.3
--- 
llvm/test/Regression/Transforms/LoopUnswitch/2006-02-14-LoopSimplifyCrash.ll:1.2
Fri Dec  1 22:23:10 2006
+++ 
llvm/test/Regression/Transforms/LoopUnswitch/2006-02-14-LoopSimplifyCrash.ll
Sun Dec 31 00:01:59 2006
@@ -1421,8 +1421,8 @@
 no_exit.53:; preds = %else.166, %else.168, %then.360, 
%no_exit.53.outer
%file.2.3.3.ph = phi int [ 0, %no_exit.53.outer ], [ %inc.551688, 
%then.360 ], [ %inc.551701, %else.168 ], [ %file.2.3.3.ph, %else.166 ]  
  ;  [#uses=2]
%nempty.5.3.ph = phi int [ 0, %no_exit.53.outer ], [ %nempty.5.3, 
%then.360 ], [ %nempty.5.3, %else.168 ], [ %nempty.5.3.ph, %else.166 ]  
  ;  [#uses=2]
-   %indvar2053 = phi uint [ 0, %no_exit.53.outer ], [ 0, %then.360 ], [ 0, 
%else.168 ], [ %indvar.next2054, %else.166 ];  [#uses=2]
-   %indvar2053 = cast uint %indvar2053 to int  ;  
[#uses=2]
+   %indvar2053.ui = phi uint [ 0, %no_exit.53.outer ], [ 0, %then.360 ], [ 
0, %else.168 ], [ %indvar.next2054, %else.166 ] ;  [#uses=2]
+   %indvar2053 = cast uint %indvar2053.ui to int   ;  
[#uses=2]
%file.2.3.3 = add int %indvar2053, %file.2.3.3.ph   ;  
[#uses=4]
%nempty.5.3 = add int %indvar2053, %nempty.5.3.ph   ;  
[#uses=3]
%tmp.4749 = add int %file.2.3.3, %tmp.4747  ;  
[#uses=1]
@@ -1472,7 +1472,7 @@
 else.166:  ; preds = %no_exit.53
%inc.55 = add int %file.2.3.3, 1;  [#uses=1]
%tmp.47421705 = setlt int %inc.55, 8;  [#uses=1]
-   %indvar.next2054 = add uint %indvar2053, 1  ;  
[#uses=1]
+   %indvar.next2054 = add uint %indvar2053.ui, 1   ;  
[#uses=1]
br bool %tmp.47421705, label %no_exit.53, label %loopexit.56
 
 loopexit.56:   ; preds = %else.166, %else.168, %then.360



___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

[llvm-commits] CVS: llvm/test/Regression/Transforms/InstCombine/2006-05-06-Infloop.ll 2006-06-28-infloop.ll 2006-09-15-CastToBool.ll 2006-10-19-SignedToUnsignedCastAndConst-2.ll 2006-10-19-SignedToUns

[Reid Spencer]

Changes in directory llvm/test/Regression/Transforms/InstCombine:

2006-05-06-Infloop.ll updated: 1.2 -> 1.3
2006-06-28-infloop.ll updated: 1.2 -> 1.3
2006-09-15-CastToBool.ll updated: 1.2 -> 1.3
2006-10-19-SignedToUnsignedCastAndConst-2.ll updated: 1.3 -> 1.4
2006-10-19-SignedToUnsignedCastAndConst.ll updated: 1.3 -> 1.4
deadcode.ll updated: 1.2 -> 1.3
fpcast.ll updated: 1.2 -> 1.3
shift-sra.ll updated: 1.4 -> 1.5
sub.ll updated: 1.23 -> 1.24
zeroext-and-reduce.ll updated: 1.2 -> 1.3
---
Log message:

For PR950: http://llvm.org/PR950 :
Update the test suite to accommodate the change from signed integer types
to signless integer types. The changes were of only a few kinds:

1. Make sure llvm-upgrade is run on the source which does the bulk of the
   changes automatically.

2. Change things like "grep 'int'" to "grep 'i32'"

3. In several tests bitcasting caused the same name to be reused in the
   same type plane. These had to be manually fixed. The fix was (generally)
   to leave the bitcast and provide the instruction with a new name. This
   should not affect the semantics of the test. In a few cases, the 
   bitcasts were known to be superfluous and irrelevant to the test case
   so they were removed.

4. One test case uses a bytecode file which needed to be updated to the
   latest bytecode format.



---
Diffs of the changes:  (+19 -29)

 2006-05-06-Infloop.ll|   18 --
 2006-06-28-infloop.ll|4 ++--
 2006-09-15-CastToBool.ll |8 
 2006-10-19-SignedToUnsignedCastAndConst-2.ll |2 +-
 2006-10-19-SignedToUnsignedCastAndConst.ll   |2 +-
 deadcode.ll  |2 +-
 fpcast.ll|6 +++---
 shift-sra.ll |2 +-
 sub.ll   |2 +-
 zeroext-and-reduce.ll|2 +-
 10 files changed, 19 insertions(+), 29 deletions(-)


Index: llvm/test/Regression/Transforms/InstCombine/2006-05-06-Infloop.ll
diff -u llvm/test/Regression/Transforms/InstCombine/2006-05-06-Infloop.ll:1.2 
llvm/test/Regression/Transforms/InstCombine/2006-05-06-Infloop.ll:1.3
--- llvm/test/Regression/Transforms/InstCombine/2006-05-06-Infloop.ll:1.2   
Fri Dec  1 22:23:09 2006
+++ llvm/test/Regression/Transforms/InstCombine/2006-05-06-Infloop.ll   Sun Dec 
31 00:01:59 2006
@@ -10,7 +10,7 @@
 
 int %mem_mono_copy_mono(%struct.gx_device* %dev, ubyte* %base, int %sourcex, 
int %raster, int %x, int %y, int %w, int %h, uint %zero, uint %one) {
 entry:
-   %raster = cast int %raster to uint  ;  [#uses=3]
+   %raster.ui = cast int %raster to uint   ;  [#uses=3]
%tmp = seteq uint %one, %zero   ;  [#uses=1]
br bool %tmp, label %cond_true, label %cond_next
 
@@ -146,7 +146,6 @@
%optr.3.2 = phi ubyte* [ %tmp232, %cond_true249 ], [ %dest.1.0, 
%cond_true249.preheader ]   ;  [#uses=1]
%bptr.3.2 = phi ubyte* [ %tmp226, %cond_true249 ], [ %line.1.0, 
%cond_true249.preheader ]   ;  [#uses=1]
%tmp. = add int %tmp109, %w ;  [#uses=1]
-   %indvar = cast uint %indvar to int  ;  [#uses=1]
%tmp.58 = mul int %indvar, -8   ;  [#uses=1]
%tmp.57 = add int %tmp., -16;  [#uses=1]
%tmp246.2 = add int %tmp.58, %tmp.57;  [#uses=1]
@@ -236,10 +235,8 @@
%dest.1.0.us = phi ubyte* [ %tmp286.us, %cond_next280.us ], [ %tmp100, 
%cond_true295.preheader.split.us ]   ;  [#uses=3]
%dest_line.1.0.us = phi ubyte** [ %tmp282.us, %cond_next280.us ], [ 
%tmp96, %cond_true295.preheader.split.us ]  ;  [#uses=1]
%tmp.89 = sub uint 0, %indvar86 ;  [#uses=1]
-   %tmp.89 = cast uint %tmp.89 to int  ;  [#uses=1]
%tmp292.0.us = add int %tmp.89, %tmp29222   ;  
[#uses=1]
-   %tmp.91 = mul uint %indvar86, %raster   ;  [#uses=1]
-   %tmp.91 = cast uint %tmp.91 to int  ;  [#uses=1]
+   %tmp.91 = mul uint %indvar86, %raster.ui;  
[#uses=1]
%tmp104.sum101 = add int %tmp102, %tmp.91   ;  
[#uses=1]
%line.1.0.us = getelementptr ubyte* %base, int %tmp104.sum101   
;  [#uses=2]
%tmp.us = load ubyte* %line.1.0.us  ;  [#uses=1]
@@ -313,10 +310,8 @@
%dest.1.0 = phi ubyte* [ %tmp286, %cond_next280 ], [ %tmp100, 
%cond_true295.preheader.split ]   ;  [#uses=4]
%dest_line.1.0 = phi ubyte** [ %tmp282, %cond_next280 ], [ %tmp96, 
%cond_true295.preheader.split ]  ;  [#uses=1]
%tmp.63 = sub uint 0, %indvar60 ;  [#uses=1]
-   %tmp.63 = cast uint %tmp.63 to int  ;  [#uses=1]
%tmp292.0 = add int %tmp.63, %tmp29222  ;  [#uses=1]
-   %tmp.65 = mul uint %indvar60, %raster   ;  [#uses=1]
-   %tmp.65

[llvm-commits] CVS: llvm/test/Regression/Analysis/LoadVN/casts.ll

[Reid Spencer]

Changes in directory llvm/test/Regression/Analysis/LoadVN:

casts.ll updated: 1.2 -> 1.3
---
Log message:

For PR950: http://llvm.org/PR950 :
Update the test suite to accommodate the change from signed integer types
to signless integer types. The changes were of only a few kinds:

1. Make sure llvm-upgrade is run on the source which does the bulk of the
   changes automatically.

2. Change things like "grep 'int'" to "grep 'i32'"

3. In several tests bitcasting caused the same name to be reused in the
   same type plane. These had to be manually fixed. The fix was (generally)
   to leave the bitcast and provide the instruction with a new name. This
   should not affect the semantics of the test. In a few cases, the 
   bitcasts were known to be superfluous and irrelevant to the test case
   so they were removed.

4. One test case uses a bytecode file which needed to be updated to the
   latest bytecode format.



---
Diffs of the changes:  (+1 -1)

 casts.ll |2 +-
 1 files changed, 1 insertion(+), 1 deletion(-)


Index: llvm/test/Regression/Analysis/LoadVN/casts.ll
diff -u llvm/test/Regression/Analysis/LoadVN/casts.ll:1.2 
llvm/test/Regression/Analysis/LoadVN/casts.ll:1.3
--- llvm/test/Regression/Analysis/LoadVN/casts.ll:1.2   Fri Dec 29 14:01:32 2006
+++ llvm/test/Regression/Analysis/LoadVN/casts.ll   Sun Dec 31 00:01:59 2006
@@ -1,7 +1,7 @@
 ; Check to make sure that Value Numbering doesn't merge casts of different
 ; flavors.
 ; RUN: llvm-upgrade < %s | llvm-as | opt -load-vn -gcse | llvm-dis | \
-; RUN: grep '[sz]ext' | wc -l | grep 2
+; RUN:   grep '[sz]ext' | wc -l | grep 2
 
 declare void %external(int)
 



___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

[llvm-commits] CVS: llvm/test/Regression/Transforms/LICM/basictest.ll no-preheader-test.ll

[Reid Spencer]

Changes in directory llvm/test/Regression/Transforms/LICM:

basictest.ll updated: 1.3 -> 1.4
no-preheader-test.ll updated: 1.3 -> 1.4
---
Log message:

For PR950: http://llvm.org/PR950 :
Update the test suite to accommodate the change from signed integer types
to signless integer types. The changes were of only a few kinds:

1. Make sure llvm-upgrade is run on the source which does the bulk of the
   changes automatically.

2. Change things like "grep 'int'" to "grep 'i32'"

3. In several tests bitcasting caused the same name to be reused in the
   same type plane. These had to be manually fixed. The fix was (generally)
   to leave the bitcast and provide the instruction with a new name. This
   should not affect the semantics of the test. In a few cases, the 
   bitcasts were known to be superfluous and irrelevant to the test case
   so they were removed.

4. One test case uses a bytecode file which needed to be updated to the
   latest bytecode format.



---
Diffs of the changes:  (+4 -4)

 basictest.ll |4 ++--
 no-preheader-test.ll |4 ++--
 2 files changed, 4 insertions(+), 4 deletions(-)


Index: llvm/test/Regression/Transforms/LICM/basictest.ll
diff -u llvm/test/Regression/Transforms/LICM/basictest.ll:1.3 
llvm/test/Regression/Transforms/LICM/basictest.ll:1.4
--- llvm/test/Regression/Transforms/LICM/basictest.ll:1.3   Fri Dec  1 
22:23:09 2006
+++ llvm/test/Regression/Transforms/LICM/basictest.ll   Sun Dec 31 00:01:59 2006
@@ -1,12 +1,12 @@
 ; RUN: llvm-upgrade < %s | llvm-as | opt -licm | llvm-dis
 
-void "testfunc"(int %i) {
+void "testfunc"(int %i.s) {
 
br label %Loop
 
 Loop:
%j = phi uint [0, %0], [%Next, %Loop]
-   %i = cast int %i to uint
+   %i = cast int %i.s to uint
%i2 = mul uint %i, 17
%Next = add uint %j, %i2
%cond = seteq uint %Next, 0


Index: llvm/test/Regression/Transforms/LICM/no-preheader-test.ll
diff -u llvm/test/Regression/Transforms/LICM/no-preheader-test.ll:1.3 
llvm/test/Regression/Transforms/LICM/no-preheader-test.ll:1.4
--- llvm/test/Regression/Transforms/LICM/no-preheader-test.ll:1.3   Fri Dec 
 1 22:23:09 2006
+++ llvm/test/Regression/Transforms/LICM/no-preheader-test.ll   Sun Dec 31 
00:01:59 2006
@@ -1,7 +1,7 @@
 ; Test that LICM works when there is not a loop-preheader
 ; RUN: llvm-upgrade < %s | llvm-as | opt -licm | llvm-dis
 
-void "testfunc"(int %i, bool %ifcond) {
+void "testfunc"(int %i.s, bool %ifcond) {
br bool %ifcond, label %Then, label %Else
 Then:
br label %Loop
@@ -10,7 +10,7 @@
 
 Loop:
%j = phi uint [0, %Then], [12, %Else], [%Next, %Loop]
-   %i = cast int %i to uint
+   %i = cast int %i.s to uint
%i2 = mul uint %i, 17
%Next = add uint %j, %i2
%cond = seteq uint %Next, 0



___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

[llvm-commits] CVS: llvm/test/Regression/CodeGen/PowerPC/2006-01-20-ShiftPartsCrash.ll and-elim.ll branch-opt.ll rotl.ll small-arguments.ll vec_spat.ll

[Reid Spencer]

Changes in directory llvm/test/Regression/CodeGen/PowerPC:

2006-01-20-ShiftPartsCrash.ll updated: 1.2 -> 1.3
and-elim.ll updated: 1.3 -> 1.4
branch-opt.ll updated: 1.3 -> 1.4
rotl.ll updated: 1.2 -> 1.3
small-arguments.ll updated: 1.5 -> 1.6
vec_spat.ll updated: 1.5 -> 1.6
---
Log message:

For PR950: http://llvm.org/PR950 :
Update the test suite to accommodate the change from signed integer types
to signless integer types. The changes were of only a few kinds:

1. Make sure llvm-upgrade is run on the source which does the bulk of the
   changes automatically.

2. Change things like "grep 'int'" to "grep 'i32'"

3. In several tests bitcasting caused the same name to be reused in the
   same type plane. These had to be manually fixed. The fix was (generally)
   to leave the bitcast and provide the instruction with a new name. This
   should not affect the semantics of the test. In a few cases, the 
   bitcasts were known to be superfluous and irrelevant to the test case
   so they were removed.

4. One test case uses a bytecode file which needed to be updated to the
   latest bytecode format.



---
Diffs of the changes:  (+33 -39)

 2006-01-20-ShiftPartsCrash.ll |   14 +++---
 and-elim.ll   |3 ++-
 branch-opt.ll |   17 -
 rotl.ll   |   20 
 small-arguments.ll|   10 --
 vec_spat.ll   |8 
 6 files changed, 33 insertions(+), 39 deletions(-)


Index: llvm/test/Regression/CodeGen/PowerPC/2006-01-20-ShiftPartsCrash.ll
diff -u llvm/test/Regression/CodeGen/PowerPC/2006-01-20-ShiftPartsCrash.ll:1.2 
llvm/test/Regression/CodeGen/PowerPC/2006-01-20-ShiftPartsCrash.ll:1.3
--- llvm/test/Regression/CodeGen/PowerPC/2006-01-20-ShiftPartsCrash.ll:1.2  
Fri Dec  1 22:23:08 2006
+++ llvm/test/Regression/CodeGen/PowerPC/2006-01-20-ShiftPartsCrash.ll  Sun Dec 
31 00:01:59 2006
@@ -1,16 +1,16 @@
 ; RUN: llvm-upgrade < %s | llvm-as | llc
 
 void %iterative_hash_host_wide_int() {
-   %zero = alloca int  ;  [#uses=2]
-   %b = alloca uint;  [#uses=1]
+   %zero = alloca int  ;  [#uses=2]
+   %b = alloca uint;  [#uses=1]
store int 0, int* %zero
-   %tmp = load int* %zero  ;  [#uses=1]
+   %tmp = load int* %zero  ;  [#uses=1]
%tmp5 = cast int %tmp to uint   ;  [#uses=1]
-   %tmp6 = add uint %tmp5, 32  ;  [#uses=1]
-   %tmp6 = cast uint %tmp6 to int  ;  [#uses=1]
-   %tmp7 = load long* null ;  [#uses=1]
+   %tmp6.u = add uint %tmp5, 32;  [#uses=1]
+   %tmp6 = cast uint %tmp6.u to int;  [#uses=1]
+   %tmp7 = load long* null ;  [#uses=1]
%tmp6 = cast int %tmp6 to ubyte ;  [#uses=1]
-   %tmp8 = shr long %tmp7, ubyte %tmp6 ;  [#uses=1]
+   %tmp8 = shr long %tmp7, ubyte %tmp6 ;  [#uses=1]
%tmp8 = cast long %tmp8 to uint ;  [#uses=1]
store uint %tmp8, uint* %b
unreachable


Index: llvm/test/Regression/CodeGen/PowerPC/and-elim.ll
diff -u llvm/test/Regression/CodeGen/PowerPC/and-elim.ll:1.3 
llvm/test/Regression/CodeGen/PowerPC/and-elim.ll:1.4
--- llvm/test/Regression/CodeGen/PowerPC/and-elim.ll:1.3Fri Dec  1 
22:23:08 2006
+++ llvm/test/Regression/CodeGen/PowerPC/and-elim.llSun Dec 31 00:01:59 2006
@@ -10,7 +10,8 @@
ret void
 }
 
-ushort %test2(ushort %crc) { ; No and's should be needed for the ushorts here.
+ushort @zext %test2(ushort @zext %crc) { 
+; No and's should be needed for the ushorts here.
 %tmp.1 = shr ushort %crc, ubyte 1
 %tmp.7 = xor ushort %tmp.1, 40961
 ret ushort %tmp.7


Index: llvm/test/Regression/CodeGen/PowerPC/branch-opt.ll
diff -u llvm/test/Regression/CodeGen/PowerPC/branch-opt.ll:1.3 
llvm/test/Regression/CodeGen/PowerPC/branch-opt.ll:1.4
--- llvm/test/Regression/CodeGen/PowerPC/branch-opt.ll:1.3  Fri Dec  1 
22:23:08 2006
+++ llvm/test/Regression/CodeGen/PowerPC/branch-opt.ll  Sun Dec 31 00:01:59 2006
@@ -9,10 +9,10 @@
 
 void %foo(int %W, int %X, int %Y, int %Z) {
 entry:
-   %X = cast int %X to uint;  [#uses=1]
-   %Y = cast int %Y to uint;  [#uses=1]
-   %Z = cast int %Z to uint;  [#uses=1]
-   %W = cast int %W to uint;  [#uses=1]
+   %X.u = cast int %X to uint  ;  [#uses=1]
+   %Y.u = cast int %Y to uint  ;  [#uses=1]
+   %Z.u = cast int %Z to uint  ;  [#uses=1]
+   %W.u = cast int %W to uint  ;  [#uses=1]
%tmp1 = and int %W, 1   ;  [#uses=1]
%tmp1 = seteq int %tmp1, 0  ;  [#uses=1]
br bool %tmp1, label %cond_false, label %bb5
@@ -21,7 +21,7 @@
%indvar77 = phi uint [ %indvar.next78, %bb ], [ 0, %bb5 ]   

[llvm-commits] CVS: llvm/test/Regression/CodeGen/ARM/vargs2.ll

[Reid Spencer]

Changes in directory llvm/test/Regression/CodeGen/ARM:

vargs2.ll updated: 1.3 -> 1.4
---
Log message:

For PR950: http://llvm.org/PR950 :
Update the test suite to accommodate the change from signed integer types
to signless integer types. The changes were of only a few kinds:

1. Make sure llvm-upgrade is run on the source which does the bulk of the
   changes automatically.

2. Change things like "grep 'int'" to "grep 'i32'"

3. In several tests bitcasting caused the same name to be reused in the
   same type plane. These had to be manually fixed. The fix was (generally)
   to leave the bitcast and provide the instruction with a new name. This
   should not affect the semantics of the test. In a few cases, the 
   bitcasts were known to be superfluous and irrelevant to the test case
   so they were removed.

4. One test case uses a bytecode file which needed to be updated to the
   latest bytecode format.



---
Diffs of the changes:  (+3 -3)

 vargs2.ll |6 +++---
 1 files changed, 3 insertions(+), 3 deletions(-)


Index: llvm/test/Regression/CodeGen/ARM/vargs2.ll
diff -u llvm/test/Regression/CodeGen/ARM/vargs2.ll:1.3 
llvm/test/Regression/CodeGen/ARM/vargs2.ll:1.4
--- llvm/test/Regression/CodeGen/ARM/vargs2.ll:1.3  Fri Dec  1 22:23:08 2006
+++ llvm/test/Regression/CodeGen/ARM/vargs2.ll  Sun Dec 31 00:01:59 2006
@@ -3,14 +3,14 @@
 
 implementation   ; Functions:
 
-void %f(int %a, ...) {
+void %f(int %a_arg, ...) {
 entry:
-   %a = cast int %a to uint;  [#uses=1]
+   %a = cast int %a_arg to uint;  [#uses=1]
%l1 = alloca sbyte*, align 4;  [#uses=5]
%l2 = alloca sbyte*, align 4;  [#uses=4]
%memtmp = alloca sbyte* ;  [#uses=2]
call void %llvm.va_start( sbyte** %l1 )
-   %tmp22 = seteq int %a, 0;  [#uses=1]
+   %tmp22 = seteq int %a_arg, 0;  [#uses=1]
%tmp23 = volatile load sbyte** %l1  ;  [#uses=2]
br bool %tmp22, label %bb8, label %bb
 



___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

[llvm-commits] CVS: llvm/test/Regression/Transforms/SimplifyLibCalls/ExitInMain.ll

[Reid Spencer]

Changes in directory llvm/test/Regression/Transforms/SimplifyLibCalls:

ExitInMain.ll updated: 1.2 -> 1.3
---
Log message:

For PR950: http://llvm.org/PR950 :
Update the test suite to accommodate the change from signed integer types
to signless integer types. The changes were of only a few kinds:

1. Make sure llvm-upgrade is run on the source which does the bulk of the
   changes automatically.

2. Change things like "grep 'int'" to "grep 'i32'"

3. In several tests bitcasting caused the same name to be reused in the
   same type plane. These had to be manually fixed. The fix was (generally)
   to leave the bitcast and provide the instruction with a new name. This
   should not affect the semantics of the test. In a few cases, the 
   bitcasts were known to be superfluous and irrelevant to the test case
   so they were removed.

4. One test case uses a bytecode file which needed to be updated to the
   latest bytecode format.



---
Diffs of the changes:  (+1 -1)

 ExitInMain.ll |2 +-
 1 files changed, 1 insertion(+), 1 deletion(-)


Index: llvm/test/Regression/Transforms/SimplifyLibCalls/ExitInMain.ll
diff -u llvm/test/Regression/Transforms/SimplifyLibCalls/ExitInMain.ll:1.2 
llvm/test/Regression/Transforms/SimplifyLibCalls/ExitInMain.ll:1.3
--- llvm/test/Regression/Transforms/SimplifyLibCalls/ExitInMain.ll:1.2  Fri Dec 
 1 22:23:10 2006
+++ llvm/test/Regression/Transforms/SimplifyLibCalls/ExitInMain.ll  Sun Dec 
31 00:02:00 2006
@@ -1,5 +1,5 @@
 ; Test that the ExitInMainOptimization pass works correctly
-; RUN: llvm-upgrade < %s | llvm-as | opt -simplify-libcalls | llvm-dis | grep 
-c 'ret int 3' | grep 1
+; RUN: llvm-upgrade < %s | llvm-as | opt -simplify-libcalls | llvm-dis | grep 
-c 'ret i32 3' | grep 1
 
 declare void %exit(int)
 declare void %exitonly(int)



___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

[llvm-commits] CVS: llvm/tools/lli/lli.cpp

[Reid Spencer]

Changes in directory llvm/tools/lli:

lli.cpp updated: 1.62 -> 1.63
---
Log message:

For PR950: http://llvm.org/PR950 :
Convert signed integer types to signless.


---
Diffs of the changes:  (+2 -2)

 lli.cpp |4 ++--
 1 files changed, 2 insertions(+), 2 deletions(-)


Index: llvm/tools/lli/lli.cpp
diff -u llvm/tools/lli/lli.cpp:1.62 llvm/tools/lli/lli.cpp:1.63
--- llvm/tools/lli/lli.cpp:1.62 Sun Dec 10 13:01:52 2006
+++ llvm/tools/lli/lli.cpp  Sun Dec 31 00:02:26 2006
@@ -121,12 +121,12 @@
 // If the program didn't explicitly call exit, call exit now, for the
 // program. This ensures that any atexit handlers get called correctly.
 Function *Exit = MP->getModule()->getOrInsertFunction("exit", Type::VoidTy,
-  Type::IntTy,
+  Type::Int32Ty,
   (Type *)0);
 
 std::vector Args;
 GenericValue ResultGV;
-ResultGV.IntVal = Result;
+ResultGV.Int32Val = Result;
 Args.push_back(ResultGV);
 EE->runFunction(Exit, Args);
 



___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

[llvm-commits] CVS: llvm/tools/llvm2cpp/CppWriter.cpp

[Reid Spencer]

Changes in directory llvm/tools/llvm2cpp:

CppWriter.cpp updated: 1.26 -> 1.27
---
Log message:

For PR950: http://llvm.org/PR950 :
Convert signed integer types to signless.


---
Diffs of the changes:  (+14 -22)

 CppWriter.cpp |   36 ++--
 1 files changed, 14 insertions(+), 22 deletions(-)


Index: llvm/tools/llvm2cpp/CppWriter.cpp
diff -u llvm/tools/llvm2cpp/CppWriter.cpp:1.26 
llvm/tools/llvm2cpp/CppWriter.cpp:1.27
--- llvm/tools/llvm2cpp/CppWriter.cpp:1.26  Sat Dec 23 00:05:41 2006
+++ llvm/tools/llvm2cpp/CppWriter.cpp   Sun Dec 31 00:02:26 2006
@@ -166,14 +166,10 @@
   switch (Ty->getTypeID()) {
 case Type::VoidTyID: prefix = "void_"; break;
 case Type::BoolTyID: prefix = "bool_"; break; 
-case Type::UByteTyID:prefix = "ubyte_"; break;
-case Type::SByteTyID:prefix = "sbyte_"; break;
-case Type::UShortTyID:   prefix = "ushort_"; break;
-case Type::ShortTyID:prefix = "short_"; break;
-case Type::UIntTyID: prefix = "uint_"; break;
-case Type::IntTyID:  prefix = "int_"; break;
-case Type::ULongTyID:prefix = "ulong_"; break;
-case Type::LongTyID: prefix = "long_"; break;
+case Type::Int8TyID: prefix = "int8_"; break;
+case Type::Int16TyID:prefix = "int16_"; break;
+case Type::Int32TyID:prefix = "int32_"; break;
+case Type::Int64TyID:prefix = "int64_"; break;
 case Type::FloatTyID:prefix = "float_"; break;
 case Type::DoubleTyID:   prefix = "double_"; break;
 case Type::LabelTyID:prefix = "label_"; break;
@@ -318,19 +314,15 @@
   // First, handle the primitive types .. easy
   if (Ty->isPrimitiveType()) {
 switch (Ty->getTypeID()) {
-  case Type::VoidTyID: return "Type::VoidTy";
-  case Type::BoolTyID: return "Type::BoolTy"; 
-  case Type::UByteTyID:return "Type::UByteTy";
-  case Type::SByteTyID:return "Type::SByteTy";
-  case Type::UShortTyID:   return "Type::UShortTy";
-  case Type::ShortTyID:return "Type::ShortTy";
-  case Type::UIntTyID: return "Type::UIntTy";
-  case Type::IntTyID:  return "Type::IntTy";
-  case Type::ULongTyID:return "Type::ULongTy";
-  case Type::LongTyID: return "Type::LongTy";
-  case Type::FloatTyID:return "Type::FloatTy";
-  case Type::DoubleTyID:   return "Type::DoubleTy";
-  case Type::LabelTyID:return "Type::LabelTy";
+  case Type::VoidTyID:   return "Type::VoidTy";
+  case Type::BoolTyID:   return "Type::BoolTy"; 
+  case Type::Int8TyID:   return "Type::Int8Ty";
+  case Type::Int16TyID:  return "Type::Int16Ty";
+  case Type::Int32TyID:  return "Type::Int32Ty";
+  case Type::Int64TyID:  return "Type::Int64Ty";
+  case Type::FloatTyID:  return "Type::FloatTy";
+  case Type::DoubleTyID: return "Type::DoubleTy";
+  case Type::LabelTyID:  return "Type::LabelTy";
   default:
 error("Invalid primitive type");
 break;
@@ -692,7 +684,7 @@
 printCFP(CFP);
 Out << ";";
   } else if (const ConstantArray *CA = dyn_cast(CV)) {
-if (CA->isString() && CA->getType()->getElementType() == Type::SByteTy) {
+if (CA->isString() && CA->getType()->getElementType() == Type::Int8Ty) {
   Out << "Constant* " << constName << " = ConstantArray::get(\"";
   printEscapedString(CA->getAsString());
   // Determine if we want null termination or not.



___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

[llvm-commits] CVS: llvm/docs/LangRef.html

[Reid Spencer]

Changes in directory llvm/docs:

LangRef.html updated: 1.180 -> 1.181
---
Log message:

For PR950: http://llvm.org/PR950 :
Update for signless integer types:
1. Replace [us]byte with i8
2. Replace [u]short with i16
3. Replace [u]int   with i32
4. Replace [u]long  with i64
5. Document the "define" keyword and use it in all examples.
6. Document parameter attributes and how they affect function types.


---
Diffs of the changes:  (+307 -277)

 LangRef.html |  584 +++
 1 files changed, 307 insertions(+), 277 deletions(-)


Index: llvm/docs/LangRef.html
diff -u llvm/docs/LangRef.html:1.180 llvm/docs/LangRef.html:1.181
--- llvm/docs/LangRef.html:1.180Thu Dec 28 10:55:55 2006
+++ llvm/docs/LangRef.html  Sun Dec 31 01:07:53 2006
@@ -24,6 +24,7 @@
   Calling Conventions
   Global Variables
   Functions
+  Parameter Attributes
   Module-Level Inline Assembly
 
   
@@ -248,7 +249,7 @@
 following instruction is syntactically okay, but not well formed:
 
 
-  %x = add int 1, %x
+  %x = add i32 1, %x
 
 
 ...because the definition of %x does not dominate all of
@@ -296,7 +297,7 @@
 ('add', 
  'bitcast', 
  'ret', etc...), for primitive type names 
('void', 'uint', etc...),
+href="#t_void">void', 'i32', 
etc...),
 and others.  These reserved words cannot conflict with variable names, because
 none of them start with a '%' character.
 
@@ -306,21 +307,21 @@
 The easy way:
 
 
-  %result = mul uint %X, 8
+  %result = mul i32 %X, 8
 
 
 After strength reduction:
 
 
-  %result = shl uint %X, ubyte 3
+  %result = shl i32 %X, i8 3
 
 
 And the hard way:
 
 
-  add uint %X, %X   ; yields {uint}:%0
-  add uint %0, %0   ; yields {uint}:%1
-  %result = add uint %1, %1
+  add i32 %X, %X   ; yields {i32}:%0
+  add i32 %0, %0   ; yields {i32}:%1
+  %result = add i32 %1, %1
 
 
 This last way of multiplying %X by 8 illustrates several
@@ -364,25 +365,25 @@
 
 ; Declare the string constant as a global constant...
 %.LC0 = internal constant [13 x sbyte] c"hello 
world\0A\00"  ; [13 x sbyte]*
+ href="#globalvars">constant [13 x i8 ] c"hello 
world\0A\00"  ; [13 x i8 ]*
 
 ; External declaration of the puts function
-declare int %puts(sbyte*) 
   ; int(sbyte*)* 
+declare i32 %puts(i8 *)   
 ; i32(i8 *)* 
 
 ; Global variable / Function body section separator
 implementation
 
 ; Definition of main function
-int %main() {; 
int()* 
-; Convert [13x sbyte]* to sbyte *...
+define i32 %main() { ; 
i32()* 
+; Convert [13x i8 ]* to i8  *...
 %cast210 = getelementptr [13 x sbyte]* %.LC0, long 0, long 0 
; sbyte*
+ href="#i_getelementptr">getelementptr [13 x i8 ]* %.LC0, i64 0, i64 0 
; i8 *
 
 ; Call puts function to write out the string to stdout...
 call int %puts(sbyte* %cast210)
  ; int
+ href="#i_call">call i32 %puts(i8 * %cast210)  
; i32
 ret int 0}
+ href="#i_ret">ret i32 0}
 
 This example is made up of a global variable
 named ".LC0", an external declaration of the "puts"
@@ -440,7 +441,7 @@
 
   "weak" linkage is exactly the same as linkonce linkage,
   except that unreferenced weak globals may not be discarded.  This is
-  used to implement constructs in C such as "int X;" at global scope.
+  used to implement constructs in C such as "i32 X;" at global scope.
   
 
   appending: 
@@ -646,14 +647,18 @@
 
 
 
-LLVM function definitions consist of an optional linkage
-type, an optional calling convention, a return
-type, a function name, a (possibly empty) argument list, an optional section,
-an optional alignment, an opening curly brace,
-a list of basic blocks, and a closing curly brace.  LLVM function declarations
-are defined with the "declare" keyword, an optional calling convention, a return type, a function name,
-a possibly empty list of arguments, and an optional alignment.
+LLVM function definitions consist of the "define" keyord, 
+an optional linkage type, an optional 
+calling convention, a return type, an optional
+parameter attribute for the return type, a function 
+name, a (possibly empty) argument list (each with optional 
+parameter attributes, an optional section, an 
optional
+alignment, an opening curly brace, a list of basic blocks, and a closing curly 
+brace.  LLVM function declarations
+are consist of the "declare" keyword, an optional calling convention, a return type, an optional
+parameter attribute for the return type, a function 
+name, a possibly empty list of arguments, and an optional alignment.
 
 A function definition contains a list of basic blocks, forming the CFG for
 the function.  Each basic block may optionally start with a label (giving the
@@ -684,6 +689,42 @@
 
 
 
+Par

[llvm-commits] CVS: llvm/docs/LangRef.html

[Reid Spencer]

Changes in directory llvm/docs:

LangRef.html updated: 1.181 -> 1.182
---
Log message:

Clean up some typos and formatting.


---
Diffs of the changes:  (+22 -20)

 LangRef.html |   42 ++
 1 files changed, 22 insertions(+), 20 deletions(-)


Index: llvm/docs/LangRef.html
diff -u llvm/docs/LangRef.html:1.181 llvm/docs/LangRef.html:1.182
--- llvm/docs/LangRef.html:1.181Sun Dec 31 01:07:53 2006
+++ llvm/docs/LangRef.html  Sun Dec 31 01:18:34 2006
@@ -652,10 +652,10 @@
 calling convention, a return type, an optional
 parameter attribute for the return type, a function 
 name, a (possibly empty) argument list (each with optional 
-parameter attributes, an optional section, an 
optional
-alignment, an opening curly brace, a list of basic blocks, and a closing curly 
-brace.  LLVM function declarations
-are consist of the "declare" keyword, an optional parameter attributes), an optional section, an 
+optional alignment, an opening curly brace, a list of basic blocks, and a 
+closing curly brace.  LLVM function declarations
+consist of the "declare" keyword, an optional calling convention, a return type, an optional
 parameter attribute for the return type, a function 
 name, a possibly empty list of arguments, and an optional alignment.
@@ -698,12 +698,12 @@
   type so two functions types that differ only by the parameter attributes 
   are different function types.
 
-  Parameter attributes consist of a at sign (@) followed by either a single
+  Parameter attributes consist of an at sign (@) followed by either a single
   keyword or a comma separate list of keywords enclosed in parentheses. For
   example:
 %someFunc = i16 @zext (i8 @(sext) %someParam)
-%someFunc = i16 @zext (i8 @zext %someParam)
-  Note that the two function types above are unique because the parameter
+%someFunc = i16 @zext (i8 @zext %someParam)
+  Note that the two function types above are unique because the parameter
   has a different attribute (@sext in the first one, @zext in the second).
 
   Currently, only the following parameter attributes are defined:
@@ -938,20 +938,22 @@
 Examples:
 
   
-
-  i32 (i32) 
-  float (i16 @sext, i32 *) *
-  i32 (i8*, ...)
+i32 (i32)
+function taking an i32, returning an i32
 
-
-  function taking an i32, returning an i32
-  Pointer to a function that takes an
-  i16 that should be sign extended and a 
+  
+float (i16 @sext, i32 *) *
+Pointer to a function that takes 
+  an i16 that should be sign extended and a 
   pointer to i32, returning 
-  float.
-  A vararg function that takes at least one pointer
-  to i8  (signed char in C), which returns an integer.  This is 
-  the signature for printf in LLVM.
+  float.
+
+  
+i32 (i8*, ...)
+A vararg function that takes at least one 
+  pointer to i8  (signed char in C), 
+  which returns an integer.  This is the signature for printf in 
+  LLVM.
 
   
 
@@ -4523,7 +4525,7 @@
 
   mailto:[EMAIL PROTECTED]">Chris Lattner
   http://llvm.org";>The LLVM Compiler Infrastructure
-  Last modified: $Date: 2006/12/31 07:07:53 $
+  Last modified: $Date: 2006/12/31 07:18:34 $
 
 
 



___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

[llvm-commits] CVS: llvm/docs/LangRef.html

[Reid Spencer]

Changes in directory llvm/docs:

LangRef.html updated: 1.182 -> 1.183
---
Log message:

One more try on the table formatting.


---
Diffs of the changes:  (+3 -2)

 LangRef.html |5 +++--
 1 files changed, 3 insertions(+), 2 deletions(-)


Index: llvm/docs/LangRef.html
diff -u llvm/docs/LangRef.html:1.182 llvm/docs/LangRef.html:1.183
--- llvm/docs/LangRef.html:1.182Sun Dec 31 01:18:34 2006
+++ llvm/docs/LangRef.html  Sun Dec 31 01:20:23 2006
@@ -942,7 +942,8 @@
 function taking an i32, returning an i32
 
   
-float (i16 @sext, i32 *) *
+float (i16 @sext, i32 *) *
+
 Pointer to a function that takes 
   an i16 that should be sign extended and a 
   pointer to i32, returning 
@@ -4525,7 +4526,7 @@
 
   mailto:[EMAIL PROTECTED]">Chris Lattner
   http://llvm.org";>The LLVM Compiler Infrastructure
-  Last modified: $Date: 2006/12/31 07:18:34 $
+  Last modified: $Date: 2006/12/31 07:20:23 $
 
 
 



___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits