summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
-rw-r--r--lib/Target/X86/X86ISelLowering.cpp21
-rw-r--r--test/CodeGen/X86/2014-05-29-factorial.ll24
2 files changed, 43 insertions, 2 deletions
diff --git a/lib/Target/X86/X86ISelLowering.cpp b/lib/Target/X86/X86ISelLowering.cpp
index e11522db79..df86064d28 100644
--- a/lib/Target/X86/X86ISelLowering.cpp
+++ b/lib/Target/X86/X86ISelLowering.cpp
@@ -10074,10 +10074,27 @@ SDValue X86TargetLowering::EmitTest(SDValue Op, unsigned X86CC, SDLoc dl,
break;
case X86::COND_G: case X86::COND_GE:
case X86::COND_L: case X86::COND_LE:
- case X86::COND_O: case X86::COND_NO:
- NeedOF = true;
+ case X86::COND_O: case X86::COND_NO: {
+ // Check if we really need to set the
+ // Overflow flag. If NoSignedWrap is present
+ // that is not actually needed.
+ switch (Op->getOpcode()) {
+ case ISD::ADD:
+ case ISD::SUB:
+ case ISD::MUL:
+ case ISD::SHL: {
+ const BinaryWithFlagsSDNode *BinNode =
+ cast<BinaryWithFlagsSDNode>(Op.getNode());
+ if (BinNode->hasNoSignedWrap())
+ break;
+ }
+ default:
+ NeedOF = true;
+ break;
+ }
break;
}
+ }
// See if we can use the EFLAGS value from the operand instead of
// doing a separate TEST. TEST always sets OF and CF to 0, so unless
// we prove that the arithmetic won't overflow, we can't use OF or CF.
diff --git a/test/CodeGen/X86/2014-05-29-factorial.ll b/test/CodeGen/X86/2014-05-29-factorial.ll
new file mode 100644
index 0000000000..70255bd321
--- /dev/null
+++ b/test/CodeGen/X86/2014-05-29-factorial.ll
@@ -0,0 +1,24 @@
+; RUN: llc < %s -march=x86-64 | FileCheck %s
+; CHECK: decq %rdi
+; CHECK-NOT: testq %rdi, %rdi
+
+define i64 @fact2(i64 %x) {
+entry:
+ br label %while.body
+
+while.body:
+ %result.06 = phi i64 [ %mul, %while.body ], [ 1, %entry ]
+ %x.addr.05 = phi i64 [ %dec, %while.body ], [ %x, %entry ]
+ %mul = mul nsw i64 %result.06, %x.addr.05
+ %dec = add nsw i64 %x.addr.05, -1
+ %cmp = icmp sgt i64 %dec, 0
+ br i1 %cmp, label %while.body, label %while.end.loopexit
+
+while.end.loopexit:
+ %mul.lcssa = phi i64 [ %mul, %while.body ]
+ br label %while.end
+
+while.end:
+ %result.0.lcssa = phi i64 [ %mul.lcssa, %while.end.loopexit ]
+ ret i64 %result.0.lcssa
+}