[X86] Turn X86ISD::AND nodes that have no flag users back into ISD::AND just before isel to enable test instruction matching

Summary:
EmitTest sometimes creates X86ISD::AND specifically to hide the AND from DAG combine. But this prevents isel patterns that look for (cmp (and X, Y), 0) from being able to see it. So we end up with an AND and a TEST. The TEST gets removed by compare instruction optimization during the peephole pass.

This patch attempts to fix this by converting X86ISD::AND with no flag users back into ISD::AND during the DAG preprocessing just before isel.

In order to do this correctly I had to make the X86ISD::AND node created by EmitTest in this case really have a flag output. Which arguably it should have had anyway so that the number of operands would be consistent for the opcode in all cases. Then I had to modify the ReplaceAllUsesWith to understand that we might be looking at an instruction with 2 outputs. Though in this case there are no uses to replace since we just created the node, but that's what the code did before so I just made it keep working.

Reviewers: spatel, RKSimon, niravd, deadalnix

Reviewed By: RKSimon

Subscribers: llvm-commits

Differential Revision: https://reviews.llvm.org/D42764

llvm-svn: 323982

5 files changed, 18 insertions, 7 deletions

diff --git a/llvm/lib/Target/X86/X86ISelDAGToDAG.cpp b/llvm/lib/Target/X86/X86ISelDAGToDAG.cpp
index 9e7de71f8a2..fb48c6466a4 100644
--- a/llvm/lib/Target/X86/X86ISelDAGToDAG.cpp
+++ b/llvm/lib/Target/X86/X86ISelDAGToDAG.cpp
@@ -625,6 +625,17 @@ void X86DAGToDAGISel::PreprocessISelDAG() {
        E = CurDAG->allnodes_end(); I != E; ) {
     SDNode *N = &*I++; // Preincrement iterator to avoid invalidation issues.
 
+    // If this is a target specific AND node with no flag usages, turn it back
+    // into ISD::AND to enable test instruction matching.
+    if (N->getOpcode() == X86ISD::AND && !N->hasAnyUseOfValue(1)) {
+      SDValue Res = CurDAG->getNode(ISD::AND, SDLoc(N), N->getValueType(0),
+                                    N->getOperand(0), N->getOperand(1));
+      --I;
+      CurDAG->ReplaceAllUsesOfValueWith(SDValue(N, 0), Res);
+      ++I;
+      CurDAG->DeleteNode(N);
+    }
+
     if (OptLevel != CodeGenOpt::None &&
         // Only do this when the target can fold the load into the call or
         // jmp.
diff --git a/llvm/lib/Target/X86/X86ISelLowering.cpp b/llvm/lib/Target/X86/X86ISelLowering.cpp
index f88b6ff7b1f..791c6dbb7e9 100644
--- a/llvm/lib/Target/X86/X86ISelLowering.cpp
+++ b/llvm/lib/Target/X86/X86ISelLowering.cpp
@@ -17365,7 +17365,8 @@ SDValue X86TargetLowering::EmitTest(SDValue Op, unsigned X86CC, const SDLoc &dl,
       if (TLI.isOperationLegal(WideVal.getOpcode(), WideVT)) {
         SDValue V0 = DAG.getNode(ISD::TRUNCATE, dl, VT, WideVal.getOperand(0));
         SDValue V1 = DAG.getNode(ISD::TRUNCATE, dl, VT, WideVal.getOperand(1));
-        Op = DAG.getNode(ConvertedOp, dl, VT, V0, V1);
+        SDVTList VTs = DAG.getVTList(Op.getValueType(), MVT::i32);
+        Op = DAG.getNode(ConvertedOp, dl, VTs, V0, V1);
       }
     }
   }
@@ -17383,7 +17384,7 @@ SDValue X86TargetLowering::EmitTest(SDValue Op, unsigned X86CC, const SDLoc &dl,
   SmallVector<SDValue, 4> Ops(Op->op_begin(), Op->op_begin() + NumOperands);
 
   SDValue New = DAG.getNode(Opcode, dl, VTs, Ops);
-  DAG.ReplaceAllUsesWith(Op, New);
+  DAG.ReplaceAllUsesOfValueWith(SDValue(Op.getNode(), 0), New);
   return SDValue(New.getNode(), 1);
 }
 
diff --git a/llvm/test/CodeGen/X86/2012-08-16-setcc.ll b/llvm/test/CodeGen/X86/2012-08-16-setcc.ll
index a31b651b3e3..f82439a7d82 100644
--- a/llvm/test/CodeGen/X86/2012-08-16-setcc.ll
+++ b/llvm/test/CodeGen/X86/2012-08-16-setcc.ll
@@ -7,7 +7,7 @@ define i32 @and_1(i8 zeroext %a, i8 zeroext %b, i32 %x) {
 ; CHECK-LABEL: and_1:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    xorl %eax, %eax
-; CHECK-NEXT:    andb %dil, %sil
+; CHECK-NEXT:    testb %dil, %sil
 ; CHECK-NEXT:    cmovnel %edx, %eax
 ; CHECK-NEXT:    retq
   %1 = and i8 %b, %a
@@ -19,7 +19,7 @@ define i32 @and_1(i8 zeroext %a, i8 zeroext %b, i32 %x) {
 define zeroext i1 @and_2(i8 zeroext %a, i8 zeroext %b) {
 ; CHECK-LABEL: and_2:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    andb %dil, %sil
+; CHECK-NEXT:    testb %dil, %sil
 ; CHECK-NEXT:    setne %al
 ; CHECK-NEXT:    retq
   %1 = and i8 %b, %a
diff --git a/llvm/test/CodeGen/X86/jump_sign.ll b/llvm/test/CodeGen/X86/jump_sign.ll
index 137edece053..5ca3b2d088d 100644
--- a/llvm/test/CodeGen/X86/jump_sign.ll
+++ b/llvm/test/CodeGen/X86/jump_sign.ll
@@ -402,8 +402,7 @@ define i32 @func_test1(i32 %p1) nounwind uwtable {
 ; CHECK-NEXT:    cmpl {{[0-9]+}}(%esp), %eax
 ; CHECK-NEXT:    setb %cl
 ; CHECK-NEXT:    movl a, %eax
-; CHECK-NEXT:    movl %eax, %edx
-; CHECK-NEXT:    andb %cl, %dl
+; CHECK-NEXT:    testb %al, %cl
 ; CHECK-NEXT:    je .LBB18_2
 ; CHECK-NEXT:  # %bb.1: # %if.then
 ; CHECK-NEXT:    decl %eax
diff --git a/llvm/test/CodeGen/X86/test-shrink-bug.ll b/llvm/test/CodeGen/X86/test-shrink-bug.ll
index a79bb0a8c21..e603c61659a 100644
--- a/llvm/test/CodeGen/X86/test-shrink-bug.ll
+++ b/llvm/test/CodeGen/X86/test-shrink-bug.ll
@@ -68,7 +68,7 @@ define void @fail(i16 %a, <2 x i8> %b) {
 ; CHECK-X64:       # %bb.0:
 ; CHECK-X64-NEXT:    pushq %rax
 ; CHECK-X64-NEXT:    .cfi_def_cfa_offset 16
-; CHECK-X64-NEXT:    andw $263, %di # imm = 0x107
+; CHECK-X64-NEXT:    testw $263, %di # imm = 0x107
 ; CHECK-X64-NEXT:    je .LBB1_2
 ; CHECK-X64-NEXT:  # %bb.1:
 ; CHECK-X64-NEXT:    pand {{.*}}(%rip), %xmm0