freebsd-dev/contrib/llvm/patches/patch-21-llvm-r221170-ppc-vaarg.diff

Pull in r221170 from upstream clang trunk (by Roman Divacky):

  Implement vaarg lowering for ppc32. Lowering of scalars and
  aggregates is supported. Complex numbers are not.

Pull in r221174 from upstream clang trunk (by Roman Divacky):

  Require asserts to unbreak the buildbots.

Pull in r221284 from upstream clang trunk (by Roman Divacky):

  Rewrite the test to not require asserts.

Pull in r221285 from upstream clang trunk (by Roman Divacky):

  Since the file has both ppc and ppc64 tests in it rename it.

This adds va_args support for PowerPC (32 bit) to clang.

Introduced here: http://svnweb.freebsd.org/changeset/base/275773

Index: tools/clang/lib/CodeGen/TargetInfo.cpp
===================================================================
--- tools/clang/lib/CodeGen/TargetInfo.cpp
+++ tools/clang/lib/CodeGen/TargetInfo.cpp
@@ -2835,12 +2835,20 @@ llvm::Value *NaClX86_64ABIInfo::EmitVAArg(llvm::Va
 
 
 // PowerPC-32
-
 namespace {
-class PPC32TargetCodeGenInfo : public DefaultTargetCodeGenInfo {
+/// PPC32_SVR4_ABIInfo - The 32-bit PowerPC ELF (SVR4) ABI information.
+class PPC32_SVR4_ABIInfo : public DefaultABIInfo {
 public:
-  PPC32TargetCodeGenInfo(CodeGenTypes &CGT) : DefaultTargetCodeGenInfo(CGT) {}
+  PPC32_SVR4_ABIInfo(CodeGen::CodeGenTypes &CGT) : DefaultABIInfo(CGT) {}
 
+  llvm::Value *EmitVAArg(llvm::Value *VAListAddr, QualType Ty,
+                         CodeGenFunction &CGF) const override;
+};
+
+class PPC32TargetCodeGenInfo : public TargetCodeGenInfo {
+public:
+  PPC32TargetCodeGenInfo(CodeGenTypes &CGT) : TargetCodeGenInfo(new PPC32_SVR4_ABIInfo(CGT)) {}
+
   int getDwarfEHStackPointer(CodeGen::CodeGenModule &M) const override {
     // This is recovered from gcc output.
     return 1; // r1 is the dedicated stack pointer
@@ -2852,6 +2860,96 @@ namespace {
 
 }
 
+llvm::Value *PPC32_SVR4_ABIInfo::EmitVAArg(llvm::Value *VAListAddr,
+                                           QualType Ty,
+                                           CodeGenFunction &CGF) const {
+  if (const ComplexType *CTy = Ty->getAs<ComplexType>()) {
+    // TODO: Implement this. For now ignore.
+    (void)CTy;
+    return nullptr;
+  }
+
+  bool isI64 = Ty->isIntegerType() && getContext().getTypeSize(Ty) == 64;
+  bool isInt = Ty->isIntegerType() || Ty->isPointerType() || Ty->isAggregateType();
+  llvm::Type *CharPtr = CGF.Int8PtrTy;
+  llvm::Type *CharPtrPtr = CGF.Int8PtrPtrTy;
+
+  CGBuilderTy &Builder = CGF.Builder;
+  llvm::Value *GPRPtr = Builder.CreateBitCast(VAListAddr, CharPtr, "gprptr");
+  llvm::Value *GPRPtrAsInt = Builder.CreatePtrToInt(GPRPtr, CGF.Int32Ty);
+  llvm::Value *FPRPtrAsInt = Builder.CreateAdd(GPRPtrAsInt, Builder.getInt32(1));
+  llvm::Value *FPRPtr = Builder.CreateIntToPtr(FPRPtrAsInt, CharPtr);
+  llvm::Value *OverflowAreaPtrAsInt = Builder.CreateAdd(FPRPtrAsInt, Builder.getInt32(3));
+  llvm::Value *OverflowAreaPtr = Builder.CreateIntToPtr(OverflowAreaPtrAsInt, CharPtrPtr);
+  llvm::Value *RegsaveAreaPtrAsInt = Builder.CreateAdd(OverflowAreaPtrAsInt, Builder.getInt32(4));
+  llvm::Value *RegsaveAreaPtr = Builder.CreateIntToPtr(RegsaveAreaPtrAsInt, CharPtrPtr);
+  llvm::Value *GPR = Builder.CreateLoad(GPRPtr, false, "gpr");
+  // Align GPR when TY is i64.
+  if (isI64) {
+    llvm::Value *GPRAnd = Builder.CreateAnd(GPR, Builder.getInt8(1));
+    llvm::Value *CC64 = Builder.CreateICmpEQ(GPRAnd, Builder.getInt8(1));
+    llvm::Value *GPRPlusOne = Builder.CreateAdd(GPR, Builder.getInt8(1));
+    GPR = Builder.CreateSelect(CC64, GPRPlusOne, GPR);
+  }
+  llvm::Value *FPR = Builder.CreateLoad(FPRPtr, false, "fpr");
+  llvm::Value *OverflowArea = Builder.CreateLoad(OverflowAreaPtr, false, "overflow_area");
+  llvm::Value *OverflowAreaAsInt = Builder.CreatePtrToInt(OverflowArea, CGF.Int32Ty);
+  llvm::Value *RegsaveArea = Builder.CreateLoad(RegsaveAreaPtr, false, "regsave_area");
+  llvm::Value *RegsaveAreaAsInt = Builder.CreatePtrToInt(RegsaveArea, CGF.Int32Ty);
+
+  llvm::Value *CC = Builder.CreateICmpULT(isInt ? GPR : FPR,
+                                          Builder.getInt8(8), "cond");
+
+  llvm::Value *RegConstant = Builder.CreateMul(isInt ? GPR : FPR,
+                                               Builder.getInt8(isInt ? 4 : 8));
+
+  llvm::Value *OurReg = Builder.CreateAdd(RegsaveAreaAsInt, Builder.CreateSExt(RegConstant, CGF.Int32Ty));
+
+  if (Ty->isFloatingType())
+    OurReg = Builder.CreateAdd(OurReg, Builder.getInt32(32));
+
+  llvm::BasicBlock *UsingRegs = CGF.createBasicBlock("using_regs");
+  llvm::BasicBlock *UsingOverflow = CGF.createBasicBlock("using_overflow");
+  llvm::BasicBlock *Cont = CGF.createBasicBlock("cont");
+
+  Builder.CreateCondBr(CC, UsingRegs, UsingOverflow);
+
+  CGF.EmitBlock(UsingRegs);
+
+  llvm::Type *PTy = llvm::PointerType::getUnqual(CGF.ConvertType(Ty));
+  llvm::Value *Result1 = Builder.CreateIntToPtr(OurReg, PTy);
+  // Increase the GPR/FPR indexes.
+  if (isInt) {
+    GPR = Builder.CreateAdd(GPR, Builder.getInt8(isI64 ? 2 : 1));
+    Builder.CreateStore(GPR, GPRPtr);
+  } else {
+    FPR = Builder.CreateAdd(FPR, Builder.getInt8(1));
+    Builder.CreateStore(FPR, FPRPtr);
+  }
+  CGF.EmitBranch(Cont);
+
+  CGF.EmitBlock(UsingOverflow);
+
+  // Increase the overflow area.
+  llvm::Value *Result2 = Builder.CreateIntToPtr(OverflowAreaAsInt, PTy);
+  OverflowAreaAsInt = Builder.CreateAdd(OverflowAreaAsInt, Builder.getInt32(isInt ? 4 : 8));
+  Builder.CreateStore(Builder.CreateIntToPtr(OverflowAreaAsInt, CharPtr), OverflowAreaPtr);
+  CGF.EmitBranch(Cont);
+
+  CGF.EmitBlock(Cont);
+
+  llvm::PHINode *Result = CGF.Builder.CreatePHI(PTy, 2, "vaarg.addr");
+  Result->addIncoming(Result1, UsingRegs);
+  Result->addIncoming(Result2, UsingOverflow);
+
+  if (Ty->isAggregateType()) {
+    llvm::Value *AGGPtr = Builder.CreateBitCast(Result, CharPtrPtr, "aggrptr")  ;
+    return Builder.CreateLoad(AGGPtr, false, "aggr");
+  }
+
+  return Result;
+}
+
 bool
 PPC32TargetCodeGenInfo::initDwarfEHRegSizeTable(CodeGen::CodeGenFunction &CGF,
                                                 llvm::Value *Address) const {
Index: tools/clang/test/CodeGen/ppc-varargs-struct.c
===================================================================
--- tools/clang/test/CodeGen/ppc-varargs-struct.c
+++ tools/clang/test/CodeGen/ppc-varargs-struct.c
@@ -0,0 +1,112 @@
+// REQUIRES: powerpc-registered-target
+// REQUIRES: asserts
+// RUN: %clang_cc1 -triple powerpc64-unknown-linux-gnu -emit-llvm -o - %s | FileCheck %s
+// RUN: %clang_cc1 -triple powerpc-unknown-linux-gnu -emit-llvm -o - %s | FileCheck %s --check-prefix=CHECK-PPC
+
+#include <stdarg.h>
+
+struct x {
+  long a;
+  double b;
+};
+
+void testva (int n, ...)
+{
+  va_list ap;
+
+  struct x t = va_arg (ap, struct x);
+// CHECK: bitcast i8* %{{[a-z.0-9]*}} to %struct.x*
+// CHECK: bitcast %struct.x* %t to i8*
+// CHECK: bitcast %struct.x* %{{[0-9]+}} to i8*
+// CHECK: call void @llvm.memcpy
+// CHECK-PPC:  [[ARRAYDECAY:%[a-z0-9]+]] = getelementptr inbounds [1 x %struct.__va_list_tag]* %ap, i32 0, i32 0
+// CHECK-PPC-NEXT:  [[GPRPTR:%[a-z0-9]+]] = bitcast %struct.__va_list_tag* [[ARRAYDECAY]] to i8*
+// CHECK-PPC-NEXT:  [[ZERO:%[0-9]+]] = ptrtoint i8* [[GPRPTR]] to i32
+// CHECK-PPC-NEXT:  [[ONE:%[0-9]+]] = add i32 [[ZERO]], 1
+// CHECK-PPC-NEXT:  [[TWO:%[0-9]+]] = inttoptr i32 [[ONE]] to i8*
+// CHECK-PPC-NEXT:  [[THREE:%[0-9]+]] = add i32 [[ONE]], 3
+// CHECK-PPC-NEXT:  [[FOUR:%[0-9]+]] = inttoptr i32 [[THREE]] to i8**
+// CHECK-PPC-NEXT:  [[FIVE:%[0-9]+]] = add i32 [[THREE]], 4
+// CHECK-PPC-NEXT:  [[SIX:%[0-9]+]] = inttoptr i32 [[FIVE]] to i8**
+// CHECK-PPC-NEXT:  [[GPR:%[a-z0-9]+]] = load i8* [[GPRPTR]]
+// CHECK-PPC-NEXT:  [[FPR:%[a-z0-9]+]] = load i8* [[TWO]] 
+// CHECK-PPC-NEXT:  [[OVERFLOW_AREA:%[a-z_0-9]+]] = load i8** [[FOUR]]
+// CHECK-PPC-NEXT:  [[SEVEN:%[0-9]+]] = ptrtoint i8* [[OVERFLOW_AREA]] to i32
+// CHECK-PPC-NEXT:  [[REGSAVE_AREA:%[a-z_0-9]+]] = load i8** [[SIX]]
+// CHECK-PPC-NEXT:  [[EIGHT:%[0-9]+]] = ptrtoint i8* [[REGSAVE_AREA]] to i32
+// CHECK-PPC-NEXT:  [[COND:%[a-z0-9]+]] = icmp ult i8 [[GPR]], 8
+// CHECK-PPC-NEXT:  [[NINE:%[0-9]+]] = mul i8 [[GPR]], 4
+// CHECK-PPC-NEXT:  [[TEN:%[0-9]+]] = sext i8 [[NINE]] to i32
+// CHECK-PPC-NEXT:  [[ELEVEN:%[0-9]+]] = add i32 [[EIGHT]], [[TEN]]
+// CHECK-PPC-NEXT:  br i1 [[COND]], label [[USING_REGS:%[a-z_0-9]+]], label [[USING_OVERFLOW:%[a-z_0-9]+]]
+//
+// CHECK-PPC1:[[USING_REGS]]
+// CHECK-PPC:  [[TWELVE:%[0-9]+]] = inttoptr i32 [[ELEVEN]] to %struct.x*
+// CHECK-PPC-NEXT:  [[THIRTEEN:%[0-9]+]] = add i8 [[GPR]], 1
+// CHECK-PPC-NEXT:  store i8 [[THIRTEEN]], i8* [[GPRPTR]]
+// CHECK-PPC-NEXT:  br label [[CONT:%[a-z0-9]+]]
+//
+// CHECK-PPC1:[[USING_OVERFLOW]]
+// CHECK-PPC:  [[FOURTEEN:%[0-9]+]] = inttoptr i32 [[SEVEN]] to %struct.x*
+// CHECK-PPC-NEXT:  [[FIFTEEN:%[0-9]+]] = add i32 [[SEVEN]], 4
+// CHECK-PPC-NEXT:  [[SIXTEEN:%[0-9]+]] = inttoptr i32 [[FIFTEEN]] to i8*
+// CHECK-PPC-NEXT:  store i8* [[SIXTEEN]], i8** [[FOUR]]
+// CHECK-PPC-NEXT:  br label [[CONT]]
+//
+// CHECK-PPC1:[[CONT]]
+// CHECK-PPC:  [[VAARG_ADDR:%[a-z.0-9]+]] = phi %struct.x* [ [[TWELVE]], [[USING_REGS]] ], [ [[FOURTEEN]], [[USING_OVERFLOW]] ]
+// CHECK-PPC-NEXT:  [[AGGRPTR:%[a-z0-9]+]] = bitcast %struct.x* [[VAARG_ADDR]] to i8**
+// CHECK-PPC-NEXT:  [[AGGR:%[a-z0-9]+]] = load i8** [[AGGRPTR]]
+// CHECK-PPC-NEXT:  [[SEVENTEEN:%[0-9]+]] = bitcast %struct.x* %t to i8*
+// CHECK-PPC-NEXT:  call void @llvm.memcpy.p0i8.p0i8.i32(i8* [[SEVENTEEN]], i8* [[AGGR]], i32 16, i32 8, i1 false)
+
+  int v = va_arg (ap, int);
+// CHECK: ptrtoint i8* %{{[a-z.0-9]*}} to i64
+// CHECK: add i64 %{{[0-9]+}}, 4
+// CHECK: inttoptr i64 %{{[0-9]+}} to i8*
+// CHECK: bitcast i8* %{{[0-9]+}} to i32*
+// CHECK-PPC:  [[ARRAYDECAY1:%[a-z0-9]+]] = getelementptr inbounds [1 x %struct.__va_list_tag]* %ap, i32 0, i32 0
+// CHECK-PPC-NEXT:  [[GPRPTR1:%[a-z0-9]+]] = bitcast %struct.__va_list_tag* [[ARRAYDECAY1]] to i8*
+// CHECK-PPC-NEXT:  [[EIGHTEEN:%[0-9]+]] = ptrtoint i8* [[GPRPTR1]] to i32
+// CHECK-PPC-NEXT:  [[NINETEEN:%[0-9]+]] = add i32 [[EIGHTEEN]], 1
+// CHECK-PPC-NEXT:  [[TWENTY:%[0-9]+]] = inttoptr i32 [[NINETEEN]] to i8*
+// CHECK-PPC-NEXT:  [[TWENTYONE:%[0-9]+]] = add i32 [[NINETEEN]], 3
+// CHECK-PPC-NEXT:  [[TWENTYTWO:%[0-9]+]] = inttoptr i32 [[TWENTYONE]] to i8**
+// CHECK-PPC-NEXT:  [[TWENTYTHREE:%[0-9]+]] = add i32 [[TWENTYONE]], 4
+// CHECK-PPC-NEXT:  [[TWENTYFOUR:%[0-9]+]] = inttoptr i32 [[TWENTYTHREE]] to i8**
+// CHECK-PPC-NEXT:  [[GPR1:%[a-z0-9]+]] = load i8* [[GPRPTR1]]
+// CHECK-PPC-NEXT:  [[FPR1:%[a-z0-9]+]] = load i8* [[TWENTY]]
+// CHECK-PPC-NEXT:  [[OVERFLOW_AREA1:%[a-z_0-9]+]] = load i8** [[TWENTYTWO]]
+// CHECK-PPC-NEXT:  [[TWENTYFIVE:%[0-9]+]] = ptrtoint i8* [[OVERFLOW_AREA1]] to i32
+// CHECK-PPC-NEXT:  [[REGSAVE_AREA1:%[a-z_0-9]+]] = load i8** [[TWENTYFOUR]]
+// CHECK-PPC-NEXT:  [[TWENTYSIX:%[0-9]+]] = ptrtoint i8* [[REGSAVE_AREA1]] to i32
+// CHECK-PPC-NEXT:  [[COND1:%[a-z0-9]+]] = icmp ult i8 [[GPR1]], 8
+// CHECK-PPC-NEXT:  [[TWENTYSEVEN:%[0-9]+]] = mul i8 [[GPR1]], 4
+// CHECK-PPC-NEXT:  [[TWENTYEIGHT:%[0-9]+]] = sext i8 [[TWENTYSEVEN]] to i32
+// CHECK-PPC-NEXT:  [[TWENTYNINE:%[0-9]+]] = add i32 [[TWENTYSIX]], [[TWENTYEIGHT]]
+// CHECK-PPC-NEXT:  br i1 [[COND1]], label [[USING_REGS1:%[a-z_0-9]+]], label [[USING_OVERFLOW1:%[a-z_0-9]+]]
+//
+// CHECK-PPC1:[[USING_REGS1]]:
+// CHECK-PPC:  [[THIRTY:%[0-9]+]] = inttoptr i32 [[TWENTYNINE]] to i32*
+// CHECK-PPC-NEXT:  [[THIRTYONE:%[0-9]+]] = add i8 [[GPR1]], 1
+// CHECK-PPC-NEXT:  store i8 [[THIRTYONE]], i8* [[GPRPTR1]]
+// CHECK-PPC-NEXT:  br label [[CONT1:%[a-z0-9]+]]
+//
+// CHECK-PPC1:[[USING_OVERFLOW1]]:
+// CHECK-PPC:  [[THIRTYTWO:%[0-9]+]] = inttoptr i32 [[TWENTYFIVE]] to i32*
+// CHECK-PPC-NEXT:  [[THIRTYTHREE:%[0-9]+]] = add i32 [[TWENTYFIVE]], 4
+// CHECK-PPC-NEXT:  [[THIRTYFOUR:%[0-9]+]] = inttoptr i32 [[THIRTYTHREE]] to i8*
+// CHECK-PPC-NEXT:  store i8* [[THIRTYFOUR]], i8** [[TWENTYTWO]]
+// CHECK-PPC-NEXT:  br label [[CONT1]]
+//
+// CHECK-PPC1:[[CONT1]]:
+// CHECK-PPC:  [[VAARG_ADDR1:%[a-z.0-9]+]] = phi i32* [ [[THIRTY]], [[USING_REGS1]] ], [ [[THIRTYTWO]], [[USING_OVERFLOW1]] ]
+// CHECK-PPC-NEXT:  [[THIRTYFIVE:%[0-9]+]] = load i32* [[VAARG_ADDR1]]
+// CHECK-PPC-NEXT:  store i32 [[THIRTYFIVE]], i32* %v, align 4
+
+#ifdef __powerpc64__
+  __int128_t u = va_arg (ap, __int128_t);
+#endif
+// CHECK: bitcast i8* %{{[a-z.0-9]+}} to i128*
+// CHECK-NEXT: load i128* %{{[0-9]+}}
+}
Index: tools/clang/test/CodeGen/ppc64-varargs-struct.c
===================================================================
--- tools/clang/test/CodeGen/ppc64-varargs-struct.c
+++ tools/clang/test/CodeGen/ppc64-varargs-struct.c
@@ -1,30 +0,0 @@
-// REQUIRES: powerpc-registered-target
-// RUN: %clang_cc1 -triple powerpc64-unknown-linux-gnu -emit-llvm -o - %s | FileCheck %s
-
-#include <stdarg.h>
-
-struct x {
-  long a;
-  double b;
-};
-
-void testva (int n, ...)
-{
-  va_list ap;
-
-  struct x t = va_arg (ap, struct x);
-// CHECK: bitcast i8* %{{[a-z.0-9]*}} to %struct.x*
-// CHECK: bitcast %struct.x* %t to i8*
-// CHECK: bitcast %struct.x* %{{[0-9]+}} to i8*
-// CHECK: call void @llvm.memcpy
-
-  int v = va_arg (ap, int);
-// CHECK: ptrtoint i8* %{{[a-z.0-9]*}} to i64
-// CHECK: add i64 %{{[0-9]+}}, 4
-// CHECK: inttoptr i64 %{{[0-9]+}} to i8*
-// CHECK: bitcast i8* %{{[0-9]+}} to i32*
-
-  __int128_t u = va_arg (ap, __int128_t);
-// CHECK: bitcast i8* %{{[a-z.0-9]+}} to i128*
-// CHECK-NEXT: load i128* %{{[0-9]+}}
-}