From 02867f0fa3e6bb1586a313a3cf4540f80538b02a Mon Sep 17 00:00:00 2001
From: Craig Topper <craig.topper@intel.com>
Date: Wed, 11 Jul 2018 06:57:42 +0000
Subject: [X86] The TEST instruction is eliminated when BSF/TZCNT is used

Summary:
These changes cover the PR#31399.
Now the ffs(x) function is lowered to (x != 0) ? llvm.cttz(x) + 1 : 0
and it corresponds to the following llvm code:
  %cnt = tail call i32 @llvm.cttz.i32(i32 %v, i1 true)
  %tobool = icmp eq i32 %v, 0
  %.op = add nuw nsw i32 %cnt, 1
  %add = select i1 %tobool, i32 0, i32 %.op
and x86 asm code:
  bsfl     %edi, %ecx
  addl     $1, %ecx
  testl    %edi, %edi
  movl     $0, %eax
  cmovnel  %ecx, %eax
In this case the 'test' instruction can't be eliminated because
the 'add' instruction modifies the EFLAGS, namely, ZF flag
that is set by the 'bsf' instruction when 'x' is zero.

We now produce the following code:
  bsfl     %edi, %ecx
  movl     $-1, %eax
  cmovnel  %ecx, %eax
  addl     $1, %eax

Patch by Ivan Kulagin

Reviewers: davide, craig.topper, spatel, RKSimon

Reviewed By: craig.topper

Subscribers: llvm-commits

Differential Revision: https://reviews.llvm.org/D48765

llvm-svn: 336768
---
 llvm/lib/Target/X86/X86ISelLowering.cpp | 30 ++++++++++++++++++++++++++++++
 1 file changed, 30 insertions(+)

(limited to 'llvm/lib/Target/X86/X86ISelLowering.cpp')

diff --git a/llvm/lib/Target/X86/X86ISelLowering.cpp b/llvm/lib/Target/X86/X86ISelLowering.cpp
index 9c74575916e..fb923436959 100644
--- a/llvm/lib/Target/X86/X86ISelLowering.cpp
+++ b/llvm/lib/Target/X86/X86ISelLowering.cpp
@@ -33452,6 +33452,36 @@ static SDValue combineCMov(SDNode *N, SelectionDAG &DAG,
     }
   }
 
+  // Handle (CMOV (ADD (CTTZ X), C), C-1, (X != 0)) ->
+  // (ADD (CMOV (CTTZ X), -1, (X != 0)), C) or
+  // (CMOV C-1, (ADD (CTTZ X), C), (X == 0)) ->
+  // (ADD (CMOV C-1, (CTTZ X), (X == 0)), C)
+  if (CC == X86::COND_NE || CC == X86::COND_E) {
+    auto *Cnst = CC == X86::COND_E ? dyn_cast<ConstantSDNode>(TrueOp)
+                                   : dyn_cast<ConstantSDNode>(FalseOp);
+    SDValue Add = CC == X86::COND_E ? FalseOp : TrueOp;
+
+    if (Cnst && Add.getOpcode() == ISD::ADD && Add.hasOneUse()) {
+      auto *AddOp1 = dyn_cast<ConstantSDNode>(Add.getOperand(1));
+      SDValue AddOp2 = Add.getOperand(0);
+      if (AddOp1 && (AddOp2.getOpcode() == ISD::CTTZ_ZERO_UNDEF ||
+                     AddOp2.getOpcode() == ISD::CTTZ)) {
+        APInt Diff = Cnst->getAPIntValue() - AddOp1->getAPIntValue();
+        if (CC == X86::COND_NE) {
+          Add = DAG.getNode(X86ISD::CMOV, DL, Add.getValueType(), AddOp2,
+                            DAG.getConstant(Diff, DL, Add.getValueType()),
+                            DAG.getConstant(CC, DL, MVT::i8), Cond);
+        } else {
+          Add = DAG.getNode(X86ISD::CMOV, DL, Add.getValueType(),
+                            DAG.getConstant(Diff, DL, Add.getValueType()),
+                            AddOp2, DAG.getConstant(CC, DL, MVT::i8), Cond);
+        }
+        return DAG.getNode(X86ISD::ADD, DL, Add.getValueType(), Add,
+                           SDValue(AddOp1, 0));
+      }
+    }
+  }
+
   return SDValue();
 }
 
-- 
cgit v1.2.3