From 32a1b257813e133ba1146289a9aa43ed8ad9c00d Mon Sep 17 00:00:00 2001
From: Anton Korobeynikov <asl@math.spbu.ru>
Date: Fri, 7 Aug 2009 22:36:50 +0000
Subject: [PATCH] 2 more vdup.32 cases

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@78419 91177308-0d34-0410-b5e6-96231b3b80d8
---
 lib/Target/ARM/ARMInstrNEON.td              | 14 ++++++++++++++
 lib/Target/ARM/AsmPrinter/ARMAsmPrinter.cpp |  5 +++++
 test/CodeGen/ARM/vdup.ll                    | 16 +++++++++++++++-
 3 files changed, 34 insertions(+), 1 deletion(-)

diff --git a/lib/Target/ARM/ARMInstrNEON.td b/lib/Target/ARM/ARMInstrNEON.td
index 7cceea22105..2c18470d754 100644
--- a/lib/Target/ARM/ARMInstrNEON.td
+++ b/lib/Target/ARM/ARMInstrNEON.td
@@ -1769,6 +1769,20 @@ def VDUPLN16q : VDUPLNQ<0b00, 0b10, "vdup.16", v8i16, v4i16>;
 def VDUPLN32q : VDUPLNQ<0b01, 0b00, "vdup.32", v4i32, v2i32>;
 def VDUPLNfq  : VDUPLNQ<0b01, 0b00, "vdup.32", v4f32, v2f32>;
 
+def VDUPfdf   : N2V<0b11, 0b11, 0b01, 0b00, 0b11000, 0, 0,
+                    (outs DPR:$dst), (ins SPR:$src),
+                    "vdup.32\t$dst, ${src:lane}", "",
+                    [(set DPR:$dst, (v2f32 (splat_lo
+                                            (scalar_to_vector SPR:$src),
+                                            undef)))]>;
+
+def VDUPfqf   : N2V<0b11, 0b11, 0b01, 0b00, 0b11000, 1, 0,
+                    (outs QPR:$dst), (ins SPR:$src),
+                    "vdup.32\t$dst, ${src:lane}", "",
+                    [(set QPR:$dst, (v4f32 (splat_lo
+                                            (scalar_to_vector SPR:$src),
+                                            undef)))]>;
+
 //   VMOVN    : Vector Narrowing Move
 defm VMOVN    : N2VNInt_HSD<0b11,0b11,0b10,0b00100,0,0, "vmovn.i",
                             int_arm_neon_vmovn>;
diff --git a/lib/Target/ARM/AsmPrinter/ARMAsmPrinter.cpp b/lib/Target/ARM/AsmPrinter/ARMAsmPrinter.cpp
index 3bbb3b9d464..c817ee932b0 100644
--- a/lib/Target/ARM/AsmPrinter/ARMAsmPrinter.cpp
+++ b/lib/Target/ARM/AsmPrinter/ARMAsmPrinter.cpp
@@ -345,6 +345,11 @@ void ARMAsmPrinter::printOperand(const MachineInstr *MI, int OpNum,
         O << '{'
           << TRI->getAsmName(DRegLo) << ',' << TRI->getAsmName(DRegHi)
           << '}';
+      } else if (Modifier && strcmp(Modifier, "lane") == 0) {
+        unsigned RegNum = ARMRegisterInfo::getRegisterNumbering(Reg);
+        unsigned DReg = TRI->getMatchingSuperReg(Reg, RegNum & 1 ? 0 : 1,
+                                                 &ARM::DPRRegClass);
+        O << TRI->getAsmName(DReg) << '[' << (RegNum & 1) << ']';
       } else {
         O << TRI->getAsmName(Reg);
       }
diff --git a/test/CodeGen/ARM/vdup.ll b/test/CodeGen/ARM/vdup.ll
index 1c0887a2492..37f8dcbaf5a 100644
--- a/test/CodeGen/ARM/vdup.ll
+++ b/test/CodeGen/ARM/vdup.ll
@@ -1,7 +1,7 @@
 ; RUN: llvm-as < %s | llc -march=arm -mattr=+neon > %t
 ; RUN: grep vdup.8 %t | count 4
 ; RUN: grep vdup.16 %t | count 4
-; RUN: grep vdup.32 %t | count 8
+; RUN: grep vdup.32 %t | count 10
 
 define <8 x i8> @v_dup8(i8 %A) nounwind {
 	%tmp1 = insertelement <8 x i8> zeroinitializer, i8 %A, i32 0
@@ -132,3 +132,17 @@ define <4 x float> @v_shuffledupQfloat(float %A) nounwind {
 	%tmp2 = shufflevector <4 x float> %tmp1, <4 x float> undef, <4 x i32> zeroinitializer
 	ret <4 x float> %tmp2
 }
+
+define <2 x float> @v_shuffledupfloat2(float* %A) nounwind {
+	%tmp0 = load float* %A
+        %tmp1 = insertelement <2 x float> undef, float %tmp0, i32 0
+        %tmp2 = shufflevector <2 x float> %tmp1, <2 x float> undef, <2 x i32> zeroinitializer
+        ret <2 x float> %tmp2
+}
+
+define <4 x float> @v_shuffledupQfloat2(float* %A) nounwind {
+        %tmp0 = load float* %A
+        %tmp1 = insertelement <4 x float> undef, float %tmp0, i32 0
+        %tmp2 = shufflevector <4 x float> %tmp1, <4 x float> undef, <4 x i32> zeroinitializer
+        ret <4 x float> %tmp2
+}