Teach dag combine to fold the following transformation more aggressively:

(OP (trunc x), (trunc y)) -> (trunc (OP x, y))

Unfortunately this simple change causes dag combine to infinite looping. The problem is the shrink demanded ops optimization tend to canonicalize expressions in the opposite manner. That is badness. This patch disable those optimizations in dag combine but instead it is done as a late pass in sdisel.

This also exposes some deficiencies in dag combine and x86 setcc / brcond lowering. Teach them to look pass ISD::TRUNCATE in various places.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@92849 91177308-0d34-0410-b5e6-96231b3b80d8

1 files changed, 13 insertions, 9 deletions

diff --git a/lib/CodeGen/SelectionDAG/TargetLowering.cpp b/lib/CodeGen/SelectionDAG/TargetLowering.cpp
index d9a5a13666..f7694dbd3e 100644
--- a/lib/CodeGen/SelectionDAG/TargetLowering.cpp
+++ b/lib/CodeGen/SelectionDAG/TargetLowering.cpp
@@ -990,7 +990,7 @@ bool TargetLowering::SimplifyDemandedBits(SDValue Op,
     if (TLO.ShrinkDemandedConstant(Op, ~KnownZero2 & NewMask))
       return true;
     // If the operation can be done in a smaller type, do so.
-    if (TLO.ShrinkDemandedOp(Op, BitWidth, NewMask, dl))
+    if (TLO.ShrinkOps && TLO.ShrinkDemandedOp(Op, BitWidth, NewMask, dl))
       return true;
 
     // Output known-1 bits are only known if set in both the LHS & RHS.
@@ -1024,7 +1024,7 @@ bool TargetLowering::SimplifyDemandedBits(SDValue Op,
     if (TLO.ShrinkDemandedConstant(Op, NewMask))
       return true;
     // If the operation can be done in a smaller type, do so.
-    if (TLO.ShrinkDemandedOp(Op, BitWidth, NewMask, dl))
+    if (TLO.ShrinkOps && TLO.ShrinkDemandedOp(Op, BitWidth, NewMask, dl))
       return true;
 
     // Output known-0 bits are only known if clear in both the LHS & RHS.
@@ -1049,7 +1049,7 @@ bool TargetLowering::SimplifyDemandedBits(SDValue Op,
     if ((KnownZero2 & NewMask) == NewMask)
       return TLO.CombineTo(Op, Op.getOperand(1));
     // If the operation can be done in a smaller type, do so.
-    if (TLO.ShrinkDemandedOp(Op, BitWidth, NewMask, dl))
+    if (TLO.ShrinkOps && TLO.ShrinkDemandedOp(Op, BitWidth, NewMask, dl))
       return true;
 
     // If all of the unknown bits are known to be zero on one side or the other
@@ -1480,7 +1480,7 @@ bool TargetLowering::SimplifyDemandedBits(SDValue Op,
                              KnownOne2, TLO, Depth+1))
       return true;
     // See if the operation should be performed at a smaller bit width.
-    if (TLO.ShrinkDemandedOp(Op, BitWidth, NewMask, dl))
+    if (TLO.ShrinkOps && TLO.ShrinkDemandedOp(Op, BitWidth, NewMask, dl))
       return true;
   }
   // FALL THROUGH
@@ -1876,7 +1876,9 @@ TargetLowering::SimplifySetCC(EVT VT, SDValue N0, SDValue N1,
 
     // Fold bit comparisons when we can.
     if ((Cond == ISD::SETEQ || Cond == ISD::SETNE) &&
-        VT == N0.getValueType() && N0.getOpcode() == ISD::AND)
+        (VT == N0.getValueType() ||
+         (isTypeLegal(VT) && VT.bitsLE(N0.getValueType()))) &&
+        N0.getOpcode() == ISD::AND)
       if (ConstantSDNode *AndRHS =
                   dyn_cast<ConstantSDNode>(N0.getOperand(1))) {
         EVT ShiftTy = DCI.isBeforeLegalize() ?
@@ -1884,16 +1886,18 @@ TargetLowering::SimplifySetCC(EVT VT, SDValue N0, SDValue N1,
         if (Cond == ISD::SETNE && C1 == 0) {// (X & 8) != 0  -->  (X & 8) >> 3
           // Perform the xform if the AND RHS is a single bit.
           if (isPowerOf2_64(AndRHS->getZExtValue())) {
-            return DAG.getNode(ISD::SRL, dl, VT, N0,
+            return DAG.getNode(ISD::TRUNCATE, dl, VT,
+                              DAG.getNode(ISD::SRL, dl, N0.getValueType(), N0,
                                 DAG.getConstant(Log2_64(AndRHS->getZExtValue()),
-                                                ShiftTy));
+                                                ShiftTy)));
           }
         } else if (Cond == ISD::SETEQ && C1 == AndRHS->getZExtValue()) {
           // (X & 8) == 8  -->  (X & 8) >> 3
           // Perform the xform if C1 is a single bit.
           if (C1.isPowerOf2()) {
-            return DAG.getNode(ISD::SRL, dl, VT, N0,
-                                DAG.getConstant(C1.logBase2(), ShiftTy));
+            return DAG.getNode(ISD::TRUNCATE, dl, VT,
+                               DAG.getNode(ISD::SRL, dl, N0.getValueType(), N0,
+                                      DAG.getConstant(C1.logBase2(), ShiftTy)));
           }
         }
       }