[llvm-commits] [llvm] r74534 - in /llvm/trunk: lib/Target/X86/X86FloatingPoint.cpp test/CodeGen/X86/inline-asm-fpstack4.ll test/CodeGen/X86/inline-asm-fpstack5.ll

[Rafael Espindola]

Author: rafael
Date: Tue Jun 30 11:40:03 2009
New Revision: 74534

URL: http://llvm.org/viewvc/llvm-project?rev=74534&view=rev
Log:
Fix PR4485.

Avoid unnecessary duplication of operand 0 of X86::FpSET_ST0_80. This duplication would
cause one register to remain on the stack at the function return.


Added:
    llvm/trunk/test/CodeGen/X86/inline-asm-fpstack5.ll
Modified:
    llvm/trunk/lib/Target/X86/X86FloatingPoint.cpp
    llvm/trunk/test/CodeGen/X86/inline-asm-fpstack4.ll

Modified: llvm/trunk/lib/Target/X86/X86FloatingPoint.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86FloatingPoint.cpp?rev=74534&r1=74533&r2=74534&view=diff

==============================================================================
--- llvm/trunk/lib/Target/X86/X86FloatingPoint.cpp (original)
+++ llvm/trunk/lib/Target/X86/X86FloatingPoint.cpp Tue Jun 30 11:40:03 2009
@@ -991,13 +991,16 @@
   case X86::FpSET_ST0_32:
   case X86::FpSET_ST0_64:
   case X86::FpSET_ST0_80: {
-    unsigned RegOnTop = getStackEntry(0);
+    unsigned Op0 = getFPReg(MI->getOperand(0));
+
     // FpSET_ST0_80 is generated by copyRegToReg for both function return
     // and inline assembly with the "st" constrain. In the latter case,
     // it is possible for ST(0) to be alive after this instruction.
-    if (!MI->killsRegister(X86::FP0 + RegOnTop)) {
-      // Duplicate ST0
+    if (!MI->killsRegister(X86::FP0 + Op0)) {
+      // Duplicate Op0
       duplicateToTop(0, 7 /*temp register*/, I);
+    } else {
+      moveToTop(Op0, I);
     }
     --StackTop;   // "Forget" we have something on the top of stack!
     break;

Modified: llvm/trunk/test/CodeGen/X86/inline-asm-fpstack4.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/inline-asm-fpstack4.ll?rev=74534&r1=74533&r2=74534&view=diff

==============================================================================
--- llvm/trunk/test/CodeGen/X86/inline-asm-fpstack4.ll (original)
+++ llvm/trunk/test/CodeGen/X86/inline-asm-fpstack4.ll Tue Jun 30 11:40:03 2009
@@ -1,4 +1,4 @@
-; RUN: llvm-as < %s | llc -march=x86 > %t
+; RUN: llvm-as < %s | llc -march=x86
 ; PR4484
 
 declare x86_fp80 @ceil()

Added: llvm/trunk/test/CodeGen/X86/inline-asm-fpstack5.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/inline-asm-fpstack5.ll?rev=74534&view=auto

==============================================================================
--- llvm/trunk/test/CodeGen/X86/inline-asm-fpstack5.ll (added)
+++ llvm/trunk/test/CodeGen/X86/inline-asm-fpstack5.ll Tue Jun 30 11:40:03 2009
@@ -0,0 +1,15 @@
+; RUN: llvm-as < %s | llc -march=x86
+; PR4485
+
+define void @test(x86_fp80* %a) {
+entry:
+	%0 = load x86_fp80* %a, align 16
+	%1 = fmul x86_fp80 %0, 0xK4006B400000000000000
+	%2 = fmul x86_fp80 %1, 0xK4012F424000000000000
+	tail call void asm sideeffect "fistpl $0", "{st},~{st}"(x86_fp80 %2)
+	%3 = load x86_fp80* %a, align 16
+	%4 = fmul x86_fp80 %3, 0xK4006B400000000000000
+	%5 = fmul x86_fp80 %4, 0xK4012F424000000000000
+	tail call void asm sideeffect "fistpl $0", "{st},~{st}"(x86_fp80 %5)
+	ret void
+}