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.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@74534 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/lib/Target/X86/X86FloatingPoint.cpp b/lib/Target/X86/X86FloatingPoint.cpp
index c282dc00947523189f8031136c7c287cc049b0c7..37027ee8bebae506808503a4afa55ded703ba386 100644 (file)
--- a/lib/Target/X86/X86FloatingPoint.cpp
+++ b/lib/Target/X86/X86FloatingPoint.cpp
@@ -991,13 +991,16 @@ void FPS::handleSpecialFP(MachineBasicBlock::iterator &I) {
   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;

diff --git a/test/CodeGen/X86/inline-asm-fpstack4.ll b/test/CodeGen/X86/inline-asm-fpstack4.ll
index 24a63537fcf8734720de3743cd1ab23140884ea1..c9122fad6cf77cbd4f5a6664c14741c3311c943d 100644 (file)
--- a/test/CodeGen/X86/inline-asm-fpstack4.ll
+++ b/test/CodeGen/X86/inline-asm-fpstack4.ll
@@ -1,4 +1,4 @@
-; RUN: llvm-as < %s | llc -march=x86 > %t
+; RUN: llvm-as < %s | llc -march=x86
 ; PR4484
 
 declare x86_fp80 @ceil()

diff --git a/test/CodeGen/X86/inline-asm-fpstack5.ll b/test/CodeGen/X86/inline-asm-fpstack5.ll
new file mode 100644 (file)
index 0000000..64f3788
--- /dev/null
+++ b/test/CodeGen/X86/inline-asm-fpstack5.ll
@@ -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
+}