X86-64:

pass/return structs of float/int as float/i32 instead of double/i64
to make the code generated for ABI cleaner.  Passing in the low part
of a double is the same as passing in a float.

For example, we now compile:

struct DeclGroup { float NumDecls; };
float foo(DeclGroup D);
void bar(DeclGroup *D) {
 foo(*D);
}

into:

%struct.DeclGroup = type { float }

define void @_Z3barP9DeclGroup(%struct.DeclGroup* %D) nounwind {
entry:
  %D.addr = alloca %struct.DeclGroup*, align 8    ; <%struct.DeclGroup**> [#uses=2]
  %agg.tmp = alloca %struct.DeclGroup, align 4    ; <%struct.DeclGroup*> [#uses=2]
  store %struct.DeclGroup* %D, %struct.DeclGroup** %D.addr
  %tmp = load %struct.DeclGroup** %D.addr         ; <%struct.DeclGroup*> [#uses=1]
  %tmp1 = bitcast %struct.DeclGroup* %agg.tmp to i8* ; <i8*> [#uses=1]
  %tmp2 = bitcast %struct.DeclGroup* %tmp to i8*  ; <i8*> [#uses=1]
  call void @llvm.memcpy.p0i8.p0i8.i64(i8* %tmp1, i8* %tmp2, i64 4, i32 4, i1 false)
  %coerce.dive = getelementptr %struct.DeclGroup* %agg.tmp, i32 0, i32 0 ; <float*> [#uses=1]
  %0 = load float* %coerce.dive, align 1          ; <float> [#uses=1]
  %call = call float @_Z3foo9DeclGroup(float %0)  ; <float> [#uses=0]
  ret void
}

instead of:

%struct.DeclGroup = type { float }

define void @_Z3barP9DeclGroup(%struct.DeclGroup* %D) nounwind {
entry:
  %D.addr = alloca %struct.DeclGroup*, align 8    ; <%struct.DeclGroup**> [#uses=2]
  %agg.tmp = alloca %struct.DeclGroup, align 4    ; <%struct.DeclGroup*> [#uses=2]
  %tmp3 = alloca double                           ; <double*> [#uses=2]
  store %struct.DeclGroup* %D, %struct.DeclGroup** %D.addr
  %tmp = load %struct.DeclGroup** %D.addr         ; <%struct.DeclGroup*> [#uses=1]
  %tmp1 = bitcast %struct.DeclGroup* %agg.tmp to i8* ; <i8*> [#uses=1]
  %tmp2 = bitcast %struct.DeclGroup* %tmp to i8*  ; <i8*> [#uses=1]
  call void @llvm.memcpy.p0i8.p0i8.i64(i8* %tmp1, i8* %tmp2, i64 4, i32 4, i1 false)
  %coerce.dive = getelementptr %struct.DeclGroup* %agg.tmp, i32 0, i32 0 ; <float*> [#uses=1]
  %0 = bitcast double* %tmp3 to float*            ; <float*> [#uses=1]
  %1 = load float* %coerce.dive                   ; <float> [#uses=1]
  store float %1, float* %0, align 1
  %2 = load double* %tmp3                         ; <double> [#uses=1]
  %call = call float @_Z3foo9DeclGroup(double %2) ; <float> [#uses=0]
  ret void
}

which is this machine code (at -O0):

__Z3barP9DeclGroup:
	subq	$24, %rsp
	movq	%rdi, 16(%rsp)
	movq	16(%rsp), %rdi
	leaq	8(%rsp), %rax
	movl	(%rdi), %ecx
	movl	%ecx, (%rax)
	movss	8(%rsp), %xmm0
	callq	__Z3foo9DeclGroup
	addq	$24, %rsp
	ret

vs this:

__Z3barP9DeclGroup:
	subq	$24, %rsp
	movq	%rdi, 16(%rsp)
	movq	16(%rsp), %rdi
	leaq	8(%rsp), %rax
	movl	(%rdi), %ecx
	movl	%ecx, (%rax)
	movss	8(%rsp), %xmm0
	movss	%xmm0, (%rsp)
	movsd	(%rsp), %xmm0
	callq	__Z3foo9DeclGroup
	addq	$24, %rsp
	ret

At -O3, it is the difference between this now:

__Z3barP9DeclGroup:
	movss	(%rdi), %xmm0
	jmp	__Z3foo9DeclGroup  # TAILCALL

vs this before:

__Z3barP9DeclGroup:
	movl	(%rdi), %eax
	movd	%rax, %xmm0
	jmp	__Z3foo9DeclGroup  # TAILCALL

llvm-svn: 107048

4 files changed, 18 insertions, 5 deletions

diff --git a/clang/lib/CodeGen/TargetInfo.cpp b/clang/lib/CodeGen/TargetInfo.cpp
index e95b3b6e901..2f8dd6c32d6 100644
--- a/clang/lib/CodeGen/TargetInfo.cpp
+++ b/clang/lib/CodeGen/TargetInfo.cpp
@@ -1086,6 +1086,13 @@ ABIArgInfo X86_64ABIInfo::getCoerceResult(QualType Ty,
     if (Ty->isIntegralOrEnumerationType() || Ty->hasPointerRepresentation())
       return (Ty->isPromotableIntegerType() ?
               ABIArgInfo::getExtend() : ABIArgInfo::getDirect());
+    
+    // If this is a 32-bit structure that is passed as an int64, then it will be
+    // passed in the low 32-bits of a 64-bit GPR, which is the same as how an
+    // i32 is passed.  Coerce to a i32 instead of a i64.
+    if (Context.getTypeSizeInChars(Ty).getQuantity() == 4)
+      CoerceTo = llvm::Type::getInt32Ty(CoerceTo->getContext());
+    
   } else if (CoerceTo->isDoubleTy()) {
     assert(Ty.isCanonical() && "should always have a canonical type here");
     assert(!Ty.hasQualifiers() && "should never have a qualified type here");
@@ -1094,6 +1101,11 @@ ABIArgInfo X86_64ABIInfo::getCoerceResult(QualType Ty,
     if (Ty == Context.FloatTy || Ty == Context.DoubleTy)
       return ABIArgInfo::getDirect();
 
+    // If this is a 32-bit structure that is passed as a double, then it will be
+    // passed in the low 32-bits of the XMM register, which is the same as how a
+    // float is passed.  Coerce to a float instead of a double.
+    if (Context.getTypeSizeInChars(Ty).getQuantity() == 4)
+      CoerceTo = llvm::Type::getFloatTy(CoerceTo->getContext());
   }
 
   return ABIArgInfo::getCoerce(CoerceTo);
diff --git a/clang/test/CodeGen/x86_64-arguments.c b/clang/test/CodeGen/x86_64-arguments.c
index d182aa25d91..24223901088 100644
--- a/clang/test/CodeGen/x86_64-arguments.c
+++ b/clang/test/CodeGen/x86_64-arguments.c
@@ -92,9 +92,10 @@ void f16(float a, float b, float c, float d, float e, float f, float g, float h,
 void f17(float a, float b, float c, float d, float e, float f, float g, float h,
          long double X) {}
 
-// Check for valid coercion.
-// CHECK: [[f18_t1:%.*]] = trunc i64 {{.*}} to i32
-// CHECK: store i32 [[f18_t1]], i32* 
+// Check for valid coercion.  The struct should be passed/returned as i32, not
+// as i64 for better code quality.
+// rdar://8135035
+// CHECK: define void @f18(i32 %a, i32) 
 struct f18_s0 { int f0; };
 void f18(int a, struct f18_s0 f18_arg1) { while (1) {} }
 
diff --git a/clang/test/CodeGenCXX/alloca-align.cpp b/clang/test/CodeGenCXX/alloca-align.cpp
index b70e366f4cf..99d6ab5845f 100644
--- a/clang/test/CodeGenCXX/alloca-align.cpp
+++ b/clang/test/CodeGenCXX/alloca-align.cpp
@@ -18,7 +18,7 @@ extern "C" void f1() {
   (void) (struct s0) { 0, 0, 0, 0 };
 }
 
-// CHECK: define i64 @f2
+// CHECK: define i32 @f2
 // CHECK: alloca %struct.s1, align 2
 struct s1 { short x; short y; };
 extern "C" struct s1 f2(int a, struct s1 *x, struct s1 *y) {
diff --git a/clang/test/CodeGenCXX/virtual-functions-incomplete-types.cpp b/clang/test/CodeGenCXX/virtual-functions-incomplete-types.cpp
index 991c2bc7220..052a0192d66 100644
--- a/clang/test/CodeGenCXX/virtual-functions-incomplete-types.cpp
+++ b/clang/test/CodeGenCXX/virtual-functions-incomplete-types.cpp
@@ -9,7 +9,7 @@ struct B {
 
 void B::f() { }
 
-// CHECK: define i64 @_ZN1D1gEv(%struct.B* %this)
+// CHECK: define i32 @_ZN1D1gEv(%struct.B* %this)
 // CHECK: declare void @_ZN1B1gEv()
 
 struct C;