Add clang patch corresponding to r275759.

This commit is contained in:
Dimitry Andric 2014-12-14 13:40:42 +00:00
parent b567315ffc
commit be038cc81f
Notes: svn2git 2020-12-20 02:59:44 +00:00
svn path=/head/; revision=275760

View File

@ -0,0 +1,241 @@
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.
This adds va_args support for PowerPC (32 bit) to clang.
Introduced here: http://svnweb.freebsd.org/changeset/base/275759
Index: tools/clang/lib/CodeGen/TargetInfo.cpp
===================================================================
--- tools/clang/lib/CodeGen/TargetInfo.cpp
+++ tools/clang/lib/CodeGen/TargetInfo.cpp
@@ -2733,12 +2733,20 @@ 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;
+};
+
+class PPC32TargetCodeGenInfo : public TargetCodeGenInfo {
+public:
+ PPC32TargetCodeGenInfo(CodeGenTypes &CGT) : TargetCodeGenInfo(new PPC32_SVR4_ABIInfo(CGT)) {}
+
int getDwarfEHStackPointer(CodeGen::CodeGenModule &M) const {
// This is recovered from gcc output.
return 1; // r1 is the dedicated stack pointer
@@ -2750,6 +2758,96 @@ 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 NULL;
+ }
+
+ 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/ppc64-varargs-struct.c
===================================================================
--- tools/clang/test/CodeGen/ppc64-varargs-struct.c
+++ tools/clang/test/CodeGen/ppc64-varargs-struct.c
@@ -1,5 +1,6 @@
// REQUIRES: ppc64-registered-target
// 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>
@@ -17,6 +18,46 @@ void testva (int n, ...)
// CHECK: bitcast %struct.x* %t to i8*
// CHECK: bitcast %struct.x* %{{[0-9]+}} to i8*
// CHECK: call void @llvm.memcpy
+// CHECK-PPC: %arraydecay = getelementptr inbounds [1 x %struct.__va_list_tag]* %ap, i32 0, i32 0
+// CHECK-PPC-NEXT: %gprptr = bitcast %struct.__va_list_tag* %arraydecay to i8*
+// CHECK-PPC-NEXT: %0 = ptrtoint i8* %gprptr to i32
+// CHECK-PPC-NEXT: %1 = add i32 %0, 1
+// CHECK-PPC-NEXT: %2 = inttoptr i32 %1 to i8*
+// CHECK-PPC-NEXT: %3 = add i32 %1, 3
+// CHECK-PPC-NEXT: %4 = inttoptr i32 %3 to i8**
+// CHECK-PPC-NEXT: %5 = add i32 %3, 4
+// CHECK-PPC-NEXT: %6 = inttoptr i32 %5 to i8**
+// CHECK-PPC-NEXT: %gpr = load i8* %gprptr
+// CHECK-PPC-NEXT: %fpr = load i8* %2
+// CHECK-PPC-NEXT: %overflow_area = load i8** %4
+// CHECK-PPC-NEXT: %7 = ptrtoint i8* %overflow_area to i32
+// CHECK-PPC-NEXT: %regsave_area = load i8** %6
+// CHECK-PPC-NEXT: %8 = ptrtoint i8* %regsave_area to i32
+// CHECK-PPC-NEXT: %cond = icmp ult i8 %gpr, 8
+// CHECK-PPC-NEXT: %9 = mul i8 %gpr, 4
+// CHECK-PPC-NEXT: %10 = sext i8 %9 to i32
+// CHECK-PPC-NEXT: %11 = add i32 %8, %10
+// CHECK-PPC-NEXT: br i1 %cond, label %using_regs, label %using_overflow
+//
+// CHECK-PPC-LABEL:using_regs: ; preds = %entry
+// CHECK-PPC-NEXT: %12 = inttoptr i32 %11 to %struct.x*
+// CHECK-PPC-NEXT: %13 = add i8 %gpr, 1
+// CHECK-PPC-NEXT: store i8 %13, i8* %gprptr
+// CHECK-PPC-NEXT: br label %cont
+//
+// CHECK-PPC-LABEL:using_overflow: ; preds = %entry
+// CHECK-PPC-NEXT: %14 = inttoptr i32 %7 to %struct.x*
+// CHECK-PPC-NEXT: %15 = add i32 %7, 4
+// CHECK-PPC-NEXT: %16 = inttoptr i32 %15 to i8*
+// CHECK-PPC-NEXT: store i8* %16, i8** %4
+// CHECK-PPC-NEXT: br label %cont
+//
+// CHECK-PPC-LABEL:cont: ; preds = %using_overflow, %using_regs
+// CHECK-PPC-NEXT: %vaarg.addr = phi %struct.x* [ %12, %using_regs ], [ %14, %using_overflow ]
+// CHECK-PPC-NEXT: %aggrptr = bitcast %struct.x* %vaarg.addr to i8**
+// CHECK-PPC-NEXT: %aggr = load i8** %aggrptr
+// CHECK-PPC-NEXT: %17 = bitcast %struct.x* %t to i8*
+// CHECK-PPC-NEXT: call void @llvm.memcpy.p0i8.p0i8.i32(i8* %17, i8* %aggr, i32 16, i32 8, i1 false)
int v = va_arg (ap, int);
// CHECK: ptrtoint i8* %{{[a-z.0-9]*}} to i64
@@ -23,8 +64,48 @@ void testva (int n, ...)
// CHECK: add i64 %{{[0-9]+}}, 4
// CHECK: inttoptr i64 %{{[0-9]+}} to i8*
// CHECK: bitcast i8* %{{[0-9]+}} to i32*
+// CHECK-PPC: %arraydecay1 = getelementptr inbounds [1 x %struct.__va_list_tag]* %ap, i32 0, i32 0
+// CHECK-PPC-NEXT: %gprptr2 = bitcast %struct.__va_list_tag* %arraydecay1 to i8*
+// CHECK-PPC-NEXT: %18 = ptrtoint i8* %gprptr2 to i32
+// CHECK-PPC-NEXT: %19 = add i32 %18, 1
+// CHECK-PPC-NEXT: %20 = inttoptr i32 %19 to i8*
+// CHECK-PPC-NEXT: %21 = add i32 %19, 3
+// CHECK-PPC-NEXT: %22 = inttoptr i32 %21 to i8**
+// CHECK-PPC-NEXT: %23 = add i32 %21, 4
+// CHECK-PPC-NEXT: %24 = inttoptr i32 %23 to i8**
+// CHECK-PPC-NEXT: %gpr3 = load i8* %gprptr2
+// CHECK-PPC-NEXT: %fpr4 = load i8* %20
+// CHECK-PPC-NEXT: %overflow_area5 = load i8** %22
+// CHECK-PPC-NEXT: %25 = ptrtoint i8* %overflow_area5 to i32
+// CHECK-PPC-NEXT: %regsave_area6 = load i8** %24
+// CHECK-PPC-NEXT: %26 = ptrtoint i8* %regsave_area6 to i32
+// CHECK-PPC-NEXT: %cond7 = icmp ult i8 %gpr3, 8
+// CHECK-PPC-NEXT: %27 = mul i8 %gpr3, 4
+// CHECK-PPC-NEXT: %28 = sext i8 %27 to i32
+// CHECK-PPC-NEXT: %29 = add i32 %26, %28
+// CHECK-PPC-NEXT: br i1 %cond7, label %using_regs8, label %using_overflow9
+//
+// CHECK-PPC-LABEL:using_regs8: ; preds = %cont
+// CHECK-PPC-NEXT: %30 = inttoptr i32 %29 to i32*
+// CHECK-PPC-NEXT: %31 = add i8 %gpr3, 1
+// CHECK-PPC-NEXT: store i8 %31, i8* %gprptr2
+// CHECK-PPC-NEXT: br label %cont10
+//
+// CHECK-PPC-LABEL:using_overflow9: ; preds = %cont
+// CHECK-PPC-NEXT: %32 = inttoptr i32 %25 to i32*
+// CHECK-PPC-NEXT: %33 = add i32 %25, 4
+// CHECK-PPC-NEXT: %34 = inttoptr i32 %33 to i8*
+// CHECK-PPC-NEXT: store i8* %34, i8** %22
+// CHECK-PPC-NEXT: br label %cont10
+//
+// CHECK-PPC-LABEL:cont10: ; preds = %using_overflow9, %using_regs8
+// CHECK-PPC-NEXT: %vaarg.addr11 = phi i32* [ %30, %using_regs8 ], [ %32, %using_overflow9 ]
+// CHECK-PPC-NEXT: %35 = load i32* %vaarg.addr11
+// CHECK-PPC-NEXT: store i32 %35, 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]+}}
}