change an assert when generating fmuladd to an ordinary 'if' check (PR25719)

[spatel]

Summary

We don't want to generate fmuladd if there's a use of the fmul expression, but this shouldn't be an assert.

The test case is derived from the commit message for r253337:
http://reviews.llvm.org/rL253337

That commit reverted r253269:
http://reviews.llvm.org/rL253269

...but the bug exists independently of the default fp-contract setting. It just became easier to hit with that change.

PR25719:
https://llvm.org/bugs/show_bug.cgi?id=25719

[scanon]

This is mostly http://reviews.llvm.org/D14891 with a test case added, but D14891 also fixed a second very minor issue: that the "else if" should just be "if". Also, majnemer made a few style suggestions there that it would be nice to adopt. Either way, we should merge the two patches. This can be the canonical one if you want to update it.

[spatel]

In D15165#300999, @scanon wrote:

This is mostly http://reviews.llvm.org/D14891 with a test case added, but D14891 also fixed a second very minor issue: that the "else if" should just be "if". Also, majnemer made a few style suggestions there that it would be nice to adopt. Either way, we should merge the two patches. This can be the canonical one if you want to update it.

Oops - too much mail and obviously missed your patch. But sure, I'll update with the suggested changes.

[spatel]

Patch updated with suggestions from D14891:

1. Changed 'else if' to plain 'if' for RHS check.
2. Used 'auto*' for dyn_casted vars.
3. Replaced 'getNUses() == 0' with 'use_empty()'.

[scanon]

LGTM.

cfe/trunk/lib/CodeGen/CGExprScalar.cpp

if (!op.FPContractable)
     return nullptr;
 
   // Check whether -ffp-contract=on. (If -ffp-contract=off/fast, fusing is
   // either disabled, or handled entirely by the LLVM backend).
   if (CGF.CGM.getCodeGenOpts().getFPContractMode() != CodeGenOptions::FPC_On)
     return nullptr;
 
   // We have a potentially fusable op. Look for a mul on one of the operands.
-  if (llvm::BinaryOperator* LHSBinOp = dyn_cast<llvm::BinaryOperator>(op.LHS)) {
-    if (LHSBinOp->getOpcode() == llvm::Instruction::FMul) {
-      assert(LHSBinOp->getNumUses() == 0 &&
-             "Operations with multiple uses shouldn't be contracted.");
+  // Also, make sure that the mul result isn't used directly. In that case,
+  // there's no point creating a muladd operation.
+  if (auto *LHSBinOp = dyn_cast<llvm::BinaryOperator>(op.LHS)) {
+    if (LHSBinOp->getOpcode() == llvm::Instruction::FMul &&
+        LHSBinOp->use_empty())
       return buildFMulAdd(LHSBinOp, op.RHS, CGF, Builder, false, isSub);
-    }
-  } else if (llvm::BinaryOperator* RHSBinOp =
-               dyn_cast<llvm::BinaryOperator>(op.RHS)) {
-    if (RHSBinOp->getOpcode() == llvm::Instruction::FMul) {
-      assert(RHSBinOp->getNumUses() == 0 &&
-             "Operations with multiple uses shouldn't be contracted.");
+  if (auto *RHSBinOp = dyn_cast<llvm::BinaryOperator>(op.RHS)) {
+    if (RHSBinOp->getOpcode() == llvm::Instruction::FMul &&
+        RHSBinOp->use_empty())
       return buildFMulAdd(RHSBinOp, op.LHS, CGF, Builder, isSub, false);
-    }
-  }
 
   return nullptr;
 }
 
 Value *ScalarExprEmitter::EmitAdd(const BinOpInfo &op) {
   if (op.LHS->getType()->isPointerTy() ||
       op.RHS->getType()->isPointerTy())
     return emitPointerArithmetic(CGF, op, /*subtraction*/ false);

cfe/trunk/test/CodeGen/fp-contract-pragma.cpp

 #pragma STDC FP_CONTRACT OFF
 float fp_contract_6(float a, float b, float c) {
 // CHECK: _Z13fp_contract_6fff
 // CHECK: %[[M:.+]] = fmul float %a, %b
 // CHECK-NEXT: fadd float %[[M]], %c
   return a * b + c;
 }
 
+// If the multiply has multiple uses, don't produce fmuladd.
+// This used to assert (PR25719):
+// https://llvm.org/bugs/show_bug.cgi?id=25719
+
+float fp_contract_7(float a, float b, float c) {
+// CHECK: _Z13fp_contract_7fff
+// CHECK:  %mul = fmul float %b, 2.000000e+00
+// CHECK-NEXT: fsub float %mul, %c
+  #pragma STDC FP_CONTRACT ON
+  return (a = 2 * b) - c;
+}
+