Make X86TargetLowering::LowerSINT_TO_FP return without creating a dead

stack slot and store if the  SINT_TO_FP is actually legal.  This allows
us to compile:

double a(double b) {return (unsigned)b;}

to:

_a:
	cvttsd2siq	%xmm0, %rax
	movl	%eax, %eax
	cvtsi2sdq	%rax, %xmm0
	ret

instead of:

_a:
	subq	$8, %rsp
	cvttsd2siq	%xmm0, %rax
	movl	%eax, %eax
	cvtsi2sdq	%rax, %xmm0
	addq	$8, %rsp
	ret

crazy.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@47660 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/lib/Target/X86/README-X86-64.txt b/lib/Target/X86/README-X86-64.txt
index efe3b73da5b1d9f4dbbc484c7500d0fec63d98d9..359b83d01a6efd88b550e3f95864117b6c483690 100644 (file)
--- a/lib/Target/X86/README-X86-64.txt
+++ b/lib/Target/X86/README-X86-64.txt
@@ -236,18 +236,3 @@ on the result of the movb).
 
 //===---------------------------------------------------------------------===//
 
-This function:
-double a(double b) {return (unsigned)b;} 
-compiles to this code:
-
-_a:
-       subq    $8, %rsp
-       cvttsd2siq      %xmm0, %rax
-       movl    %eax, %eax
-       cvtsi2sdq       %rax, %xmm0
-       addq    $8, %rsp
-       ret
-
-note the dead rsp adjustments.
-
-//===---------------------------------------------------------------------===//

diff --git a/lib/Target/X86/X86ISelLowering.cpp b/lib/Target/X86/X86ISelLowering.cpp
index df30aa545ea8564a99365cd98c0f215be6a0563c..50014b31824a1a4c4b9e5c28bbc769d4e6fdea3b 100644 (file)
--- a/lib/Target/X86/X86ISelLowering.cpp
+++ b/lib/Target/X86/X86ISelLowering.cpp
@@ -4148,12 +4148,17 @@ SDOperand X86TargetLowering::LowerShift(SDOperand Op, SelectionDAG &DAG) {
 }
 
 SDOperand X86TargetLowering::LowerSINT_TO_FP(SDOperand Op, SelectionDAG &DAG) {
-  assert(Op.getOperand(0).getValueType() <= MVT::i64 &&
-         Op.getOperand(0).getValueType() >= MVT::i16 &&
-         "Unknown SINT_TO_FP to lower!");
-
-  SDOperand Result;
   MVT::ValueType SrcVT = Op.getOperand(0).getValueType();
+  assert(SrcVT <= MVT::i64 && SrcVT >= MVT::i16 &&
+         "Unknown SINT_TO_FP to lower!");
+  
+  // These are really Legal; caller falls through into that case.
+  if (SrcVT == MVT::i32 && isScalarFPTypeInSSEReg(Op.getValueType()))
+    return SDOperand();
+  if (SrcVT == MVT::i64 && Op.getValueType() != MVT::f80 && 
+      Subtarget->is64Bit())
+    return SDOperand();
+  
   unsigned Size = MVT::getSizeInBits(SrcVT)/8;
   MachineFunction &MF = DAG.getMachineFunction();
   int SSFI = MF.getFrameInfo()->CreateStackObject(Size, Size);
@@ -4163,13 +4168,6 @@ SDOperand X86TargetLowering::LowerSINT_TO_FP(SDOperand Op, SelectionDAG &DAG) {
                                  PseudoSourceValue::getFixedStack(),
                                  SSFI);
 
-  // These are really Legal; caller falls through into that case.
-  if (SrcVT == MVT::i32 && isScalarFPTypeInSSEReg(Op.getValueType()))
-    return Result;
-  if (SrcVT == MVT::i64 && Op.getValueType() != MVT::f80 && 
-      Subtarget->is64Bit())
-    return Result;
-
   // Build the FILD
   SDVTList Tys;
   bool useSSE = isScalarFPTypeInSSEReg(Op.getValueType());
@@ -4181,8 +4179,8 @@ SDOperand X86TargetLowering::LowerSINT_TO_FP(SDOperand Op, SelectionDAG &DAG) {
   Ops.push_back(Chain);
   Ops.push_back(StackSlot);
   Ops.push_back(DAG.getValueType(SrcVT));
-  Result = DAG.getNode(useSSE ? X86ISD::FILD_FLAG :X86ISD::FILD,
-                       Tys, &Ops[0], Ops.size());
+  SDOperand Result = DAG.getNode(useSSE ? X86ISD::FILD_FLAG : X86ISD::FILD,
+                                 Tys, &Ops[0], Ops.size());
 
   if (useSSE) {
     Chain = Result.getValue(1);

diff --git a/test/CodeGen/X86/x86-64-dead-stack-adjust.ll b/test/CodeGen/X86/x86-64-dead-stack-adjust.ll
new file mode 100644 (file)
index 0000000..9f95e41
--- /dev/null
+++ b/test/CodeGen/X86/x86-64-dead-stack-adjust.ll
@@ -0,0 +1,10 @@
+; RUN: llvm-as < %s | not grep rsp
+target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128"
+target triple = "x86_64-apple-darwin8"
+
+define double @a(double %b) nounwind  {
+entry:
+       %tmp12 = fptoui double %b to i32                ; <i32> [#uses=1]
+       %tmp123 = uitofp i32 %tmp12 to double           ; <double> [#uses=1]
+       ret double %tmp123
+}