From 637cc6a8aa9dc5280449c703c263d10bfaba3172 Mon Sep 17 00:00:00 2001
From: Chandler Carruth <chandlerc@gmail.com>
Date: Tue, 13 Dec 2011 01:56:10 +0000
Subject: Initial CodeGen support for CTTZ/CTLZ where a zero input produces an
 undefined result. This adds new ISD nodes for the new semantics, selecting
 them when the LLVM intrinsic indicates that the undef behavior is desired.
 The new nodes expand trivially to the old nodes, so targets don't actually
 need to do anything to support these new nodes besides indicating that they
 should be expanded. I've done this for all the operand types that I could
 figure out for all the targets. Owners of various targets, please review and
 let me know if any of these are incorrect.

Note that the expand behavior is *conservatively correct*, and exactly
matches LLVM's current behavior with these operations. Ideally this
patch will not change behavior in any way. For example the regtest suite
finds the exact same instruction sequences coming out of the code
generator. That's why there are no new tests here -- all of this is
being exercised by the existing test suite.

Thanks to Duncan Sands for reviewing the various bits of this patch and
helping me get the wrinkles ironed out with expanding for each target.
Also thanks to Chris for clarifying through all the discussions that
this is indeed the approach he was looking for. That said, there are
likely still rough spots. Further review much appreciated.

llvm-svn: 146466
---
 llvm/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp | 18 +++++++++++++++---
 1 file changed, 15 insertions(+), 3 deletions(-)

(limited to 'llvm/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp')

diff --git a/llvm/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp b/llvm/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp
index 0230a85bd52..75f57611126 100644
--- a/llvm/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp
+++ b/llvm/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp
@@ -2382,6 +2382,9 @@ SDValue SelectionDAGLegalize::ExpandBitCount(unsigned Opc, SDValue Op,
 
     return Op;
   }
+  case ISD::CTLZ_ZERO_UNDEF:
+    // This trivially expands to CTLZ.
+    return DAG.getNode(ISD::CTLZ, dl, Op.getValueType(), Op);
   case ISD::CTLZ: {
     // for now, we do this:
     // x = x | (x >> 1);
@@ -2403,6 +2406,9 @@ SDValue SelectionDAGLegalize::ExpandBitCount(unsigned Opc, SDValue Op,
     Op = DAG.getNOT(dl, Op, VT);
     return DAG.getNode(ISD::CTPOP, dl, VT, Op);
   }
+  case ISD::CTTZ_ZERO_UNDEF:
+    // This trivially expands to CTTZ.
+    return DAG.getNode(ISD::CTTZ, dl, Op.getValueType(), Op);
   case ISD::CTTZ: {
     // for now, we use: { return popcount(~x & (x - 1)); }
     // unless the target has ctlz but not ctpop, in which case we use:
@@ -2517,7 +2523,9 @@ void SelectionDAGLegalize::ExpandNode(SDNode *Node) {
   switch (Node->getOpcode()) {
   case ISD::CTPOP:
   case ISD::CTLZ:
+  case ISD::CTLZ_ZERO_UNDEF:
   case ISD::CTTZ:
+  case ISD::CTTZ_ZERO_UNDEF:
     Tmp1 = ExpandBitCount(Node->getOpcode(), Node->getOperand(0), dl);
     Results.push_back(Tmp1);
     break;
@@ -3419,20 +3427,24 @@ void SelectionDAGLegalize::PromoteNode(SDNode *Node) {
   SDValue Tmp1, Tmp2, Tmp3;
   switch (Node->getOpcode()) {
   case ISD::CTTZ:
+  case ISD::CTTZ_ZERO_UNDEF:
   case ISD::CTLZ:
+  case ISD::CTLZ_ZERO_UNDEF:
   case ISD::CTPOP:
     // Zero extend the argument.
     Tmp1 = DAG.getNode(ISD::ZERO_EXTEND, dl, NVT, Node->getOperand(0));
-    // Perform the larger operation.
+    // Perform the larger operation. For CTPOP and CTTZ_ZERO_UNDEF, this is
+    // already the correct result.
     Tmp1 = DAG.getNode(Node->getOpcode(), dl, NVT, Tmp1);
     if (Node->getOpcode() == ISD::CTTZ) {
-      //if Tmp1 == sizeinbits(NVT) then Tmp1 = sizeinbits(Old VT)
+      // FIXME: This should set a bit in the zero extended value instead.
       Tmp2 = DAG.getSetCC(dl, TLI.getSetCCResultType(NVT),
                           Tmp1, DAG.getConstant(NVT.getSizeInBits(), NVT),
                           ISD::SETEQ);
       Tmp1 = DAG.getNode(ISD::SELECT, dl, NVT, Tmp2,
                           DAG.getConstant(OVT.getSizeInBits(), NVT), Tmp1);
-    } else if (Node->getOpcode() == ISD::CTLZ) {
+    } else if (Node->getOpcode() == ISD::CTLZ ||
+               Node->getOpcode() == ISD::CTLZ_ZERO_UNDEF) {
       // Tmp1 = Tmp1 - (sizeinbits(NVT) - sizeinbits(Old VT))
       Tmp1 = DAG.getNode(ISD::SUB, dl, NVT, Tmp1,
                           DAG.getConstant(NVT.getSizeInBits() -
-- 
cgit v1.2.3