1
0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-10-19 19:12:56 +02:00

Changed SSE4/AVX <2 x i64> extract and insert ops to be Custom lowered

Constant idx case is still done in tablegen but other cases are then expanded

Fixes <rdar://problem/10435460>

llvm-svn: 144557
This commit is contained in:
Pete Cooper 2011-11-14 19:38:42 +00:00
parent 720b712c77
commit c9d6834f38
2 changed files with 34 additions and 5 deletions

View File

@ -944,9 +944,11 @@ X86TargetLowering::X86TargetLowering(X86TargetMachine &TM)
setOperationAction(ISD::EXTRACT_VECTOR_ELT, MVT::v4i32, Custom);
setOperationAction(ISD::EXTRACT_VECTOR_ELT, MVT::v4f32, Custom);
// FIXME: these should be Legal but thats only for the case where
// the index is constant. For now custom expand to deal with that
if (Subtarget->is64Bit()) {
setOperationAction(ISD::INSERT_VECTOR_ELT, MVT::v2i64, Legal);
setOperationAction(ISD::EXTRACT_VECTOR_ELT, MVT::v2i64, Legal);
setOperationAction(ISD::INSERT_VECTOR_ELT, MVT::v2i64, Custom);
setOperationAction(ISD::EXTRACT_VECTOR_ELT, MVT::v2i64, Custom);
}
}
@ -6963,8 +6965,8 @@ X86TargetLowering::LowerEXTRACT_VECTOR_ELT_SSE4(SDValue Op,
Op.getOperand(0)),
Op.getOperand(1));
return DAG.getNode(ISD::BITCAST, dl, MVT::f32, Extract);
} else if (VT == MVT::i32) {
// ExtractPS works with constant index.
} else if (VT == MVT::i32 || VT == MVT::i64) {
// ExtractPS/pextrq works with constant index.
if (isa<ConstantSDNode>(Op.getOperand(1)))
return Op;
}
@ -7103,7 +7105,8 @@ X86TargetLowering::LowerINSERT_VECTOR_ELT_SSE4(SDValue Op,
// Create this as a scalar to vector..
N1 = DAG.getNode(ISD::SCALAR_TO_VECTOR, dl, MVT::v4f32, N1);
return DAG.getNode(X86ISD::INSERTPS, dl, VT, N0, N1, N2);
} else if (EltVT == MVT::i32 && isa<ConstantSDNode>(N2)) {
} else if ((EltVT == MVT::i32 || EltVT == MVT::i64) &&
isa<ConstantSDNode>(N2)) {
// PINSR* works with constant index.
return Op;
}

View File

@ -0,0 +1,26 @@
; RUN: llc < %s -march=x86-64 -mattr=+sse41
target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64"
target triple = "x86_64-apple-darwin11.0.0"
define i64 @__builtin_ia32_vec_ext_v2di(<2 x i64> %a, i32 %i) nounwind {
%1 = alloca <2 x i64>, align 16
%2 = alloca i32, align 4
store <2 x i64> %a, <2 x i64>* %1, align 16
store i32 %i, i32* %2, align 4
%3 = load <2 x i64>* %1, align 16
%4 = load i32* %2, align 4
%5 = extractelement <2 x i64> %3, i32 %4
ret i64 %5
}
define <2 x i64> @__builtin_ia32_vec_int_v2di(<2 x i64> %a, i32 %i) nounwind {
%1 = alloca <2 x i64>, align 16
%2 = alloca i32, align 4
store <2 x i64> %a, <2 x i64>* %1, align 16
store i32 %i, i32* %2, align 4
%3 = load <2 x i64>* %1, align 16
%4 = load i32* %2, align 4
%5 = insertelement <2 x i64> %3, i64 1, i32 %4
ret <2 x i64> %5
}