From 807f9cf2433ad59b080dae153de034ad58d8f3e1 Mon Sep 17 00:00:00 2001 From: Simon Pilgrim Date: Fri, 11 Nov 2016 11:51:29 +0000 Subject: [PATCH] [SelectionDAG] Add support for vector demandedelts in BSWAP opcodes llvm-svn: 286582 --- llvm/lib/CodeGen/SelectionDAG/SelectionDAG.cpp | 3 ++- llvm/test/CodeGen/X86/known-bits-vector.ll | 8 ++------ 2 files changed, 4 insertions(+), 7 deletions(-) diff --git a/llvm/lib/CodeGen/SelectionDAG/SelectionDAG.cpp b/llvm/lib/CodeGen/SelectionDAG/SelectionDAG.cpp index eee4cda..7a9dd57 100644 --- a/llvm/lib/CodeGen/SelectionDAG/SelectionDAG.cpp +++ b/llvm/lib/CodeGen/SelectionDAG/SelectionDAG.cpp @@ -2575,7 +2575,8 @@ void SelectionDAG::computeKnownBits(SDValue Op, APInt &KnownZero, break; } case ISD::BSWAP: { - computeKnownBits(Op.getOperand(0), KnownZero2, KnownOne2, Depth+1); + computeKnownBits(Op.getOperand(0), KnownZero2, KnownOne2, DemandedElts, + Depth + 1); KnownZero = KnownZero2.byteSwap(); KnownOne = KnownOne2.byteSwap(); break; diff --git a/llvm/test/CodeGen/X86/known-bits-vector.ll b/llvm/test/CodeGen/X86/known-bits-vector.ll index 0cdd1f3..068894b 100644 --- a/llvm/test/CodeGen/X86/known-bits-vector.ll +++ b/llvm/test/CodeGen/X86/known-bits-vector.ll @@ -309,16 +309,12 @@ define <4 x i32> @knownbits_mask_srem_shuffle_lshr(<4 x i32> %a0) nounwind { define <4 x i32> @knownbits_mask_bswap_shuffle_shl(<4 x i32> %a0) nounwind { ; X32-LABEL: knownbits_mask_bswap_shuffle_shl: ; X32: # BB#0: -; X32-NEXT: vpand {{\.LCPI.*}}, %xmm0, %xmm0 -; X32-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[3,2,1,0,3,2,1,0,15,14,13,12,15,14,13,12] -; X32-NEXT: vpslld $22, %xmm0, %xmm0 +; X32-NEXT: vxorps %xmm0, %xmm0, %xmm0 ; X32-NEXT: retl ; ; X64-LABEL: knownbits_mask_bswap_shuffle_shl: ; X64: # BB#0: -; X64-NEXT: vpand {{.*}}(%rip), %xmm0, %xmm0 -; X64-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[3,2,1,0,3,2,1,0,15,14,13,12,15,14,13,12] -; X64-NEXT: vpslld $22, %xmm0, %xmm0 +; X64-NEXT: vxorps %xmm0, %xmm0, %xmm0 ; X64-NEXT: retq %1 = and <4 x i32> %a0, %2 = call <4 x i32> @llvm.bswap.v4i32(<4 x i32> %1) -- 2.7.4