From: Simon Pilgrim <llvm-dev@redking.me.uk>
Date: Mon, 30 Oct 2017 19:31:08 +0000 (+0000)
Subject: [SelectionDAG] Add VSELECT demanded elts support to computeKnownBits
X-Git-Tag: llvmorg-6.0.0-rc1~4565
X-Git-Url: http://review.tizen.org/git/?a=commitdiff_plain;h=017f896adb9d41702d0727e7e008917a8a8d5bc3;p=platform%2Fupstream%2Fllvm.git

[SelectionDAG] Add VSELECT demanded elts support to computeKnownBits

llvm-svn: 316947
---

diff --git a/llvm/lib/CodeGen/SelectionDAG/SelectionDAG.cpp b/llvm/lib/CodeGen/SelectionDAG/SelectionDAG.cpp
index a7b19ee..16d157a 100644
--- a/llvm/lib/CodeGen/SelectionDAG/SelectionDAG.cpp
+++ b/llvm/lib/CodeGen/SelectionDAG/SelectionDAG.cpp
@@ -2417,22 +2417,22 @@ void SelectionDAG::computeKnownBits(SDValue Op, KnownBits &Known,
   }
   case ISD::SELECT:
   case ISD::VSELECT:
-    computeKnownBits(Op.getOperand(2), Known, Depth+1);
+    computeKnownBits(Op.getOperand(2), Known, DemandedElts, Depth+1);
     // If we don't know any bits, early out.
     if (Known.isUnknown())
       break;
-    computeKnownBits(Op.getOperand(1), Known2, Depth+1);
+    computeKnownBits(Op.getOperand(1), Known2, DemandedElts, Depth+1);
 
     // Only known if known in both the LHS and RHS.
     Known.One &= Known2.One;
     Known.Zero &= Known2.Zero;
     break;
   case ISD::SELECT_CC:
-    computeKnownBits(Op.getOperand(3), Known, Depth+1);
+    computeKnownBits(Op.getOperand(3), Known, DemandedElts, Depth+1);
     // If we don't know any bits, early out.
     if (Known.isUnknown())
       break;
-    computeKnownBits(Op.getOperand(2), Known2, Depth+1);
+    computeKnownBits(Op.getOperand(2), Known2, DemandedElts, Depth+1);
 
     // Only known if known in both the LHS and RHS.
     Known.One &= Known2.One;
diff --git a/llvm/test/CodeGen/X86/known-bits-vector.ll b/llvm/test/CodeGen/X86/known-bits-vector.ll
index 67c0b6a..e9b2d67 100644
--- a/llvm/test/CodeGen/X86/known-bits-vector.ll
+++ b/llvm/test/CodeGen/X86/known-bits-vector.ll
@@ -618,11 +618,7 @@ define <4 x float> @knownbits_and_select_shuffle_uitofp(<4 x i32> %a0, <4 x i32>
 ; X32-NEXT:    vpcmpeqd %xmm1, %xmm0, %xmm0
 ; X32-NEXT:    vblendvps %xmm0, %xmm2, %xmm3, %xmm0
 ; X32-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[0,0,2,2]
-; X32-NEXT:    vpblendw {{.*#+}} xmm1 = xmm0[0],mem[1],xmm0[2],mem[3],xmm0[4],mem[5],xmm0[6],mem[7]
-; X32-NEXT:    vpsrld $16, %xmm0, %xmm0
-; X32-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0],mem[1],xmm0[2],mem[3],xmm0[4],mem[5],xmm0[6],mem[7]
-; X32-NEXT:    vaddps {{\.LCPI.*}}, %xmm0, %xmm0
-; X32-NEXT:    vaddps %xmm0, %xmm1, %xmm0
+; X32-NEXT:    vcvtdq2ps %xmm0, %xmm0
 ; X32-NEXT:    movl %ebp, %esp
 ; X32-NEXT:    popl %ebp
 ; X32-NEXT:    retl
@@ -634,11 +630,7 @@ define <4 x float> @knownbits_and_select_shuffle_uitofp(<4 x i32> %a0, <4 x i32>
 ; X64-NEXT:    vpcmpeqd %xmm1, %xmm0, %xmm0
 ; X64-NEXT:    vblendvps %xmm0, %xmm2, %xmm3, %xmm0
 ; X64-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[0,0,2,2]
-; X64-NEXT:    vpblendw {{.*#+}} xmm1 = xmm0[0],mem[1],xmm0[2],mem[3],xmm0[4],mem[5],xmm0[6],mem[7]
-; X64-NEXT:    vpsrld $16, %xmm0, %xmm0
-; X64-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0],mem[1],xmm0[2],mem[3],xmm0[4],mem[5],xmm0[6],mem[7]
-; X64-NEXT:    vaddps {{.*}}(%rip), %xmm0, %xmm0
-; X64-NEXT:    vaddps %xmm0, %xmm1, %xmm0
+; X64-NEXT:    vcvtdq2ps %xmm0, %xmm0
 ; X64-NEXT:    retq
   %1 = and <4 x i32> %a2, <i32 65535, i32 -1, i32 255, i32 -1>
   %2 = and <4 x i32> %a3, <i32 255, i32 -1, i32 65535, i32 -1>
@@ -664,11 +656,7 @@ define <4 x float> @knownbits_lshr_and_select_shuffle_uitofp(<4 x i32> %a0, <4 x
 ; X32-NEXT:    vpcmpeqd %xmm1, %xmm0, %xmm0
 ; X32-NEXT:    vblendvps %xmm0, %xmm2, %xmm3, %xmm0
 ; X32-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[0,0,2,2]
-; X32-NEXT:    vpblendw {{.*#+}} xmm1 = xmm0[0],mem[1],xmm0[2],mem[3],xmm0[4],mem[5],xmm0[6],mem[7]
-; X32-NEXT:    vpsrld $16, %xmm0, %xmm0
-; X32-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0],mem[1],xmm0[2],mem[3],xmm0[4],mem[5],xmm0[6],mem[7]
-; X32-NEXT:    vaddps {{\.LCPI.*}}, %xmm0, %xmm0
-; X32-NEXT:    vaddps %xmm0, %xmm1, %xmm0
+; X32-NEXT:    vcvtdq2ps %xmm0, %xmm0
 ; X32-NEXT:    movl %ebp, %esp
 ; X32-NEXT:    popl %ebp
 ; X32-NEXT:    retl
@@ -682,11 +670,7 @@ define <4 x float> @knownbits_lshr_and_select_shuffle_uitofp(<4 x i32> %a0, <4 x
 ; X64-NEXT:    vpcmpeqd %xmm1, %xmm0, %xmm0
 ; X64-NEXT:    vblendvps %xmm0, %xmm2, %xmm3, %xmm0
 ; X64-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[0,0,2,2]
-; X64-NEXT:    vpblendw {{.*#+}} xmm1 = xmm0[0],mem[1],xmm0[2],mem[3],xmm0[4],mem[5],xmm0[6],mem[7]
-; X64-NEXT:    vpsrld $16, %xmm0, %xmm0
-; X64-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0],mem[1],xmm0[2],mem[3],xmm0[4],mem[5],xmm0[6],mem[7]
-; X64-NEXT:    vaddps {{.*}}(%rip), %xmm0, %xmm0
-; X64-NEXT:    vaddps %xmm0, %xmm1, %xmm0
+; X64-NEXT:    vcvtdq2ps %xmm0, %xmm0
 ; X64-NEXT:    retq
   %1 = lshr <4 x i32> %a2, <i32 5, i32 1, i32 5, i32 1>
   %2 = and <4 x i32> %a3, <i32 255, i32 -1, i32 65535, i32 -1>