[llvm-branch-commits] [llvm] 76ceebb - [DAGCombine] Check the uses of negated floating constant and remove the hack

Tom Stellard via llvm-branch-commits llvm-branch-commits at lists.llvm.org
Fri Jun 26 13:46:57 PDT 2020


Author: QingShan Zhang
Date: 2020-06-26T13:45:41-07:00
New Revision: 76ceebb0d96361de2b61bf8e504d306d1f2e492f

URL: https://github.com/llvm/llvm-project/commit/76ceebb0d96361de2b61bf8e504d306d1f2e492f
DIFF: https://github.com/llvm/llvm-project/commit/76ceebb0d96361de2b61bf8e504d306d1f2e492f.diff

LOG: [DAGCombine] Check the uses of negated floating constant and remove the hack

PowerPC hits an assertion due to somewhat the same reason as https://reviews.llvm.org/D70975.
Though there are already some hack, it still failed with some case, when the operand 0 is NOT
a const fp, it is another fma that with const fp. And that const fp is negated which result in multi-uses.

A better fix is to check the uses of the negated const fp. If there are already use of its negated
value, we will have benefit as no extra Node is added.

Differential revision: https://reviews.llvm.org/D75501

(cherry picked from commit 3906ae387f0775dfe4426e4336748269fafbd190)

Added: 
    

Modified: 
    llvm/lib/CodeGen/SelectionDAG/TargetLowering.cpp
    llvm/test/CodeGen/PowerPC/fma-combine.ll

Removed: 
    


################################################################################
diff  --git a/llvm/lib/CodeGen/SelectionDAG/TargetLowering.cpp b/llvm/lib/CodeGen/SelectionDAG/TargetLowering.cpp
index 24ab65171a17..368e2100031f 100644
--- a/llvm/lib/CodeGen/SelectionDAG/TargetLowering.cpp
+++ b/llvm/lib/CodeGen/SelectionDAG/TargetLowering.cpp
@@ -5490,9 +5490,20 @@ char TargetLowering::isNegatibleForFree(SDValue Op, SelectionDAG &DAG,
   EVT VT = Op.getValueType();
   const SDNodeFlags Flags = Op->getFlags();
   const TargetOptions &Options = DAG.getTarget().Options;
-  if (!Op.hasOneUse() && !(Op.getOpcode() == ISD::FP_EXTEND &&
-                           isFPExtFree(VT, Op.getOperand(0).getValueType())))
-    return 0;
+  if (!Op.hasOneUse()) {
+    bool IsFreeExtend = Op.getOpcode() == ISD::FP_EXTEND &&
+                        isFPExtFree(VT, Op.getOperand(0).getValueType());
+
+    // If we already have the use of the negated floating constant, it is free
+    // to negate it even it has multiple uses.
+    bool IsFreeConstant =
+        Op.getOpcode() == ISD::ConstantFP &&
+        !getNegatedExpression(Op, DAG, LegalOperations, ForCodeSize)
+             .use_empty();
+
+    if (!IsFreeExtend && !IsFreeConstant)
+      return 0;
+  }
 
   // Don't recurse exponentially.
   if (Depth > SelectionDAG::MaxRecursionDepth)
@@ -5687,14 +5698,7 @@ SDValue TargetLowering::getNegatedExpression(SDValue Op, SelectionDAG &DAG,
                                  ForCodeSize, Depth + 1);
     char V1 = isNegatibleForFree(Op.getOperand(1), DAG, LegalOperations,
                                  ForCodeSize, Depth + 1);
-    // TODO: This is a hack. It is possible that costs have changed between now
-    //       and the initial calls to isNegatibleForFree(). That is because we
-    //       are rewriting the expression, and that may change the number of
-    //       uses (and therefore the cost) of values. If the negation costs are
-    //       equal, only negate this value if it is a constant. Otherwise, try
-    //       operand 1. A better fix would eliminate uses as a cost factor or
-    //       track the change in uses as we rewrite the expression.
-    if (V0 > V1 || (V0 == V1 && isa<ConstantFPSDNode>(Op.getOperand(0)))) {
+    if (V0 > V1) {
       // fold (fneg (fma X, Y, Z)) -> (fma (fneg X), Y, (fneg Z))
       SDValue Neg0 = getNegatedExpression(
           Op.getOperand(0), DAG, LegalOperations, ForCodeSize, Depth + 1);

diff  --git a/llvm/test/CodeGen/PowerPC/fma-combine.ll b/llvm/test/CodeGen/PowerPC/fma-combine.ll
index 88da295201fe..8973dc1334d7 100644
--- a/llvm/test/CodeGen/PowerPC/fma-combine.ll
+++ b/llvm/test/CodeGen/PowerPC/fma-combine.ll
@@ -137,3 +137,104 @@ entry:
   %add = fsub double %mul, %a
   ret double %add
 }
+
+define float @fma_combine_no_ice() {
+; CHECK-FAST-LABEL: fma_combine_no_ice:
+; CHECK-FAST:       # %bb.0:
+; CHECK-FAST-NEXT:    addis 3, 2, .LCPI4_0 at toc@ha
+; CHECK-FAST-NEXT:    addis 4, 2, .LCPI4_1 at toc@ha
+; CHECK-FAST-NEXT:    lfs 0, .LCPI4_0 at toc@l(3)
+; CHECK-FAST-NEXT:    lfsx 2, 0, 3
+; CHECK-FAST-NEXT:    addis 3, 2, .LCPI4_2 at toc@ha
+; CHECK-FAST-NEXT:    lfs 3, .LCPI4_1 at toc@l(4)
+; CHECK-FAST-NEXT:    lfs 1, .LCPI4_2 at toc@l(3)
+; CHECK-FAST-NEXT:    xsmaddasp 3, 2, 0
+; CHECK-FAST-NEXT:    xsmaddasp 1, 2, 3
+; CHECK-FAST-NEXT:    xsnmsubasp 1, 3, 2
+; CHECK-FAST-NEXT:    blr
+;
+; CHECK-FAST-NOVSX-LABEL: fma_combine_no_ice:
+; CHECK-FAST-NOVSX:       # %bb.0:
+; CHECK-FAST-NOVSX-NEXT:    addis 3, 2, .LCPI4_0 at toc@ha
+; CHECK-FAST-NOVSX-NEXT:    lfs 0, .LCPI4_0 at toc@l(3)
+; CHECK-FAST-NOVSX-NEXT:    addis 3, 2, .LCPI4_1 at toc@ha
+; CHECK-FAST-NOVSX-NEXT:    lfs 1, 0(3)
+; CHECK-FAST-NOVSX-NEXT:    lfs 2, .LCPI4_1 at toc@l(3)
+; CHECK-FAST-NOVSX-NEXT:    addis 3, 2, .LCPI4_2 at toc@ha
+; CHECK-FAST-NOVSX-NEXT:    fmadds 0, 1, 2, 0
+; CHECK-FAST-NOVSX-NEXT:    lfs 2, .LCPI4_2 at toc@l(3)
+; CHECK-FAST-NOVSX-NEXT:    fmadds 2, 1, 0, 2
+; CHECK-FAST-NOVSX-NEXT:    fnmsubs 1, 0, 1, 2
+; CHECK-FAST-NOVSX-NEXT:    blr
+;
+; CHECK-LABEL: fma_combine_no_ice:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addis 3, 2, .LCPI4_0 at toc@ha
+; CHECK-NEXT:    addis 4, 2, .LCPI4_1 at toc@ha
+; CHECK-NEXT:    lfs 0, .LCPI4_0 at toc@l(3)
+; CHECK-NEXT:    lfsx 2, 0, 3
+; CHECK-NEXT:    addis 3, 2, .LCPI4_2 at toc@ha
+; CHECK-NEXT:    lfs 3, .LCPI4_1 at toc@l(4)
+; CHECK-NEXT:    lfs 1, .LCPI4_2 at toc@l(3)
+; CHECK-NEXT:    xsmaddasp 3, 2, 0
+; CHECK-NEXT:    xsmaddasp 1, 2, 3
+; CHECK-NEXT:    xsnmsubasp 1, 3, 2
+; CHECK-NEXT:    blr
+  %tmp = load float, float* undef, align 4
+  %tmp2 = load float, float* undef, align 4
+  %tmp3 = fmul fast float %tmp, 0x3FE372D780000000
+  %tmp4 = fadd fast float %tmp3, 1.000000e+00
+  %tmp5 = fmul fast float %tmp2, %tmp4
+  %tmp6 = load float, float* undef, align 4
+  %tmp7 = load float, float* undef, align 4
+  %tmp8 = fmul fast float %tmp7, 0x3FE372D780000000
+  %tmp9 = fsub fast float -1.000000e+00, %tmp8
+  %tmp10 = fmul fast float %tmp9, %tmp6
+  %tmp11 = fadd fast float %tmp5, 5.000000e-01
+  %tmp12 = fadd fast float %tmp11, %tmp10
+  ret float %tmp12
+}
+
+; This would crash while trying getNegatedExpression().
+define double @getNegatedExpression_crash(double %x, double %y) {
+; CHECK-FAST-LABEL: getNegatedExpression_crash:
+; CHECK-FAST:       # %bb.0:
+; CHECK-FAST-NEXT:    addis 3, 2, .LCPI5_1 at toc@ha
+; CHECK-FAST-NEXT:    addis 4, 2, .LCPI5_0 at toc@ha
+; CHECK-FAST-NEXT:    lfs 3, .LCPI5_1 at toc@l(3)
+; CHECK-FAST-NEXT:    lfs 4, .LCPI5_0 at toc@l(4)
+; CHECK-FAST-NEXT:    xssubdp 0, 1, 3
+; CHECK-FAST-NEXT:    xsmaddadp 3, 1, 4
+; CHECK-FAST-NEXT:    xsmaddadp 0, 3, 2
+; CHECK-FAST-NEXT:    fmr 1, 0
+; CHECK-FAST-NEXT:    blr
+;
+; CHECK-FAST-NOVSX-LABEL: getNegatedExpression_crash:
+; CHECK-FAST-NOVSX:       # %bb.0:
+; CHECK-FAST-NOVSX-NEXT:    addis 3, 2, .LCPI5_0 at toc@ha
+; CHECK-FAST-NOVSX-NEXT:    addis 4, 2, .LCPI5_1 at toc@ha
+; CHECK-FAST-NOVSX-NEXT:    lfs 0, .LCPI5_0 at toc@l(3)
+; CHECK-FAST-NOVSX-NEXT:    lfs 3, .LCPI5_1 at toc@l(4)
+; CHECK-FAST-NOVSX-NEXT:    fmadd 3, 1, 3, 0
+; CHECK-FAST-NOVSX-NEXT:    fsub 0, 1, 0
+; CHECK-FAST-NOVSX-NEXT:    fmadd 1, 3, 2, 0
+; CHECK-FAST-NOVSX-NEXT:    blr
+;
+; CHECK-LABEL: getNegatedExpression_crash:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addis 3, 2, .LCPI5_1 at toc@ha
+; CHECK-NEXT:    addis 4, 2, .LCPI5_0 at toc@ha
+; CHECK-NEXT:    lfs 3, .LCPI5_1 at toc@l(3)
+; CHECK-NEXT:    lfs 4, .LCPI5_0 at toc@l(4)
+; CHECK-NEXT:    xssubdp 0, 1, 3
+; CHECK-NEXT:    xsmaddadp 3, 1, 4
+; CHECK-NEXT:    xsmaddadp 0, 3, 2
+; CHECK-NEXT:    fmr 1, 0
+; CHECK-NEXT:    blr
+  %neg = fneg fast double %x
+  %fma = call fast double @llvm.fma.f64(double %neg, double 42.0, double -1.0)
+  %add = fadd fast double %x, 1.0
+  %fma1 = call fast double @llvm.fma.f64(double %fma, double %y, double %add)
+  ret double %fma1
+}
+declare double @llvm.fma.f64(double, double, double) nounwind readnone


        


More information about the llvm-branch-commits mailing list