[llvm] r286071 - [SelectionDAG] Add support for vector demandedelts in OR opcodes

[Simon Pilgrim via llvm-commits]

Author: rksimon
Date: Sun Nov  6 10:29:09 2016
New Revision: 286071

URL: http://llvm.org/viewvc/llvm-project?rev=286071&view=rev
Log:
[SelectionDAG] Add support for vector demandedelts in OR opcodes

Modified:
    llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAG.cpp
    llvm/trunk/test/CodeGen/X86/known-bits-vector.ll

Modified: llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAG.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAG.cpp?rev=286071&r1=286070&r2=286071&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAG.cpp (original)
+++ llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAG.cpp Sun Nov  6 10:29:09 2016
@@ -2142,8 +2142,10 @@ void SelectionDAG::computeKnownBits(SDVa
     KnownZero |= KnownZero2;
     break;
   case ISD::OR:
-    computeKnownBits(Op.getOperand(1), KnownZero, KnownOne, Depth+1);
-    computeKnownBits(Op.getOperand(0), KnownZero2, KnownOne2, Depth+1);
+    computeKnownBits(Op.getOperand(1), KnownZero, KnownOne, DemandedElts,
+                     Depth + 1);
+    computeKnownBits(Op.getOperand(0), KnownZero2, KnownOne2, DemandedElts,
+                     Depth + 1);
 
     // Output known-0 bits are only known if clear in both the LHS & RHS.
     KnownZero &= KnownZero2;

Modified: llvm/trunk/test/CodeGen/X86/known-bits-vector.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/known-bits-vector.ll?rev=286071&r1=286070&r2=286071&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/X86/known-bits-vector.ll (original)
+++ llvm/trunk/test/CodeGen/X86/known-bits-vector.ll Sun Nov  6 10:29:09 2016
@@ -96,11 +96,7 @@ define <4 x float> @knownbits_mask_or_sh
 ; X32-NEXT:    vpand {{\.LCPI.*}}, %xmm0, %xmm0
 ; X32-NEXT:    vpor {{\.LCPI.*}}, %xmm0, %xmm0
 ; X32-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,2,3,3]
-; 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:    retl
 ;
 ; X64-LABEL: knownbits_mask_or_shuffle_uitofp:
@@ -108,11 +104,7 @@ define <4 x float> @knownbits_mask_or_sh
 ; X64-NEXT:    vpand {{.*}}(%rip), %xmm0, %xmm0
 ; X64-NEXT:    vpor {{.*}}(%rip), %xmm0, %xmm0
 ; X64-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,2,3,3]
-; 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> %a0, <i32 -1, i32 -1, i32 255, i32 4085>
   %2 = or <4 x i32> %1, <i32 65535, i32 65535, i32 65535, i32 65535>