From 9d50559bae511cd75ea61efb7189e4b954ab4175 Mon Sep 17 00:00:00 2001
From: Owen Anderson <resistor@mac.com>
Date: Thu, 21 Oct 2010 18:20:25 +0000
Subject: [PATCH] Add correct encodings for NEON vaddw.s* and vaddw.u*.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@117040 91177308-0d34-0410-b5e6-96231b3b80d8
---
 lib/Target/ARM/ARMInstrNEON.td  | 22 +++++++++---
 test/MC/ARM/neon-fp-encoding.ll | 60 +++++++++++++++++++++++++++++++++
 2 files changed, 78 insertions(+), 4 deletions(-)

diff --git a/lib/Target/ARM/ARMInstrNEON.td b/lib/Target/ARM/ARMInstrNEON.td
index 7a63085a9d0..45d10e3a505 100644
--- a/lib/Target/ARM/ARMInstrNEON.td
+++ b/lib/Target/ARM/ARMInstrNEON.td
@@ -1668,11 +1668,25 @@ class N3VW<bit op24, bit op23, bits<2> op21_20, bits<4> op11_8, bit op4,
            string OpcodeStr, string Dt, ValueType TyQ, ValueType TyD,
            SDNode OpNode, SDNode ExtOp, bit Commutable>
   : N3V<op24, op23, op21_20, op11_8, 0, op4,
-        (outs QPR:$dst), (ins QPR:$src1, DPR:$src2), N3RegFrm, IIC_VSUBiD,
-        OpcodeStr, Dt, "$dst, $src1, $src2", "",
-        [(set QPR:$dst, (OpNode (TyQ QPR:$src1),
-                                (TyQ (ExtOp (TyD DPR:$src2)))))]> {
+        (outs QPR:$Qd), (ins QPR:$Qn, DPR:$Dm), N3RegFrm, IIC_VSUBiD,
+        OpcodeStr, Dt, "$Qd, $Qn, $Dm", "",
+        [(set QPR:$Qd, (OpNode (TyQ QPR:$Qn),
+                                (TyQ (ExtOp (TyD DPR:$Dm)))))]> {
   let isCommutable = Commutable;
+  
+  // Instruction operands.
+  bits<4> Qd;
+  bits<4> Qn;
+  bits<5> Dm;
+  
+  let Inst{15-13} = Qd{2-0};
+  let Inst{22}    = Qd{3};
+  let Inst{12}    = 0;
+  let Inst{19-17} = Qn{2-0};
+  let Inst{7}     = Qn{3};
+  let Inst{16}    = 0;
+  let Inst{3-0}   = Dm{3-0};
+  let Inst{5}     = Dm{4};
 }
 
 // Pairwise long 2-register intrinsics, both double- and quad-register.
diff --git a/test/MC/ARM/neon-fp-encoding.ll b/test/MC/ARM/neon-fp-encoding.ll
index 52880364339..9cd5bd77008 100644
--- a/test/MC/ARM/neon-fp-encoding.ll
+++ b/test/MC/ARM/neon-fp-encoding.ll
@@ -119,3 +119,63 @@ define <2 x i64> @vaddlu_2xi32(<2 x i32>* %A, <2 x i32>* %B) nounwind {
 	%tmp5 = add <2 x i64> %tmp3, %tmp4
 	ret <2 x i64> %tmp5
 }
+
+; CHECK: vaddws_8xi8
+define <8 x i16> @vaddws_8xi8(<8 x i16>* %A, <8 x i8>* %B) nounwind {
+	%tmp1 = load <8 x i16>* %A
+	%tmp2 = load <8 x i8>* %B
+	%tmp3 = sext <8 x i8> %tmp2 to <8 x i16>
+; CHECK: vaddw.s8	q8, q8, d18     @ encoding: [0xa2,0x01,0xc0,0xf2]
+	%tmp4 = add <8 x i16> %tmp1, %tmp3
+	ret <8 x i16> %tmp4
+}
+
+; CHECK: vaddws_4xi16
+define <4 x i32> @vaddws_4xi16(<4 x i32>* %A, <4 x i16>* %B) nounwind {
+	%tmp1 = load <4 x i32>* %A
+	%tmp2 = load <4 x i16>* %B
+	%tmp3 = sext <4 x i16> %tmp2 to <4 x i32>
+; CHECK: vaddw.s16	q8, q8, d18     @ encoding: [0xa2,0x01,0xd0,0xf2]
+	%tmp4 = add <4 x i32> %tmp1, %tmp3
+	ret <4 x i32> %tmp4
+}
+
+; CHECK: vaddws_2xi32
+define <2 x i64> @vaddws_2xi32(<2 x i64>* %A, <2 x i32>* %B) nounwind {
+	%tmp1 = load <2 x i64>* %A
+	%tmp2 = load <2 x i32>* %B
+	%tmp3 = sext <2 x i32> %tmp2 to <2 x i64>
+; CHECK: vaddw.s32	q8, q8, d18     @ encoding: [0xa2,0x01,0xe0,0xf2]
+	%tmp4 = add <2 x i64> %tmp1, %tmp3
+	ret <2 x i64> %tmp4
+}
+
+; CHECK: vaddwu_8xi8
+define <8 x i16> @vaddwu_8xi8(<8 x i16>* %A, <8 x i8>* %B) nounwind {
+	%tmp1 = load <8 x i16>* %A
+	%tmp2 = load <8 x i8>* %B
+	%tmp3 = zext <8 x i8> %tmp2 to <8 x i16>
+; CHECK: vaddw.u8	q8, q8, d18     @ encoding: [0xa2,0x01,0xc0,0xf3]
+	%tmp4 = add <8 x i16> %tmp1, %tmp3
+	ret <8 x i16> %tmp4
+}
+
+; CHECK: vaddwu_4xi16
+define <4 x i32> @vaddwu_4xi16(<4 x i32>* %A, <4 x i16>* %B) nounwind {
+	%tmp1 = load <4 x i32>* %A
+	%tmp2 = load <4 x i16>* %B
+	%tmp3 = zext <4 x i16> %tmp2 to <4 x i32>
+; CHECK: vaddw.u16	q8, q8, d18     @ encoding: [0xa2,0x01,0xd0,0xf3]
+	%tmp4 = add <4 x i32> %tmp1, %tmp3
+	ret <4 x i32> %tmp4
+}
+
+; CHECK: vaddwu_2xi32
+define <2 x i64> @vaddwu_2xi32(<2 x i64>* %A, <2 x i32>* %B) nounwind {
+	%tmp1 = load <2 x i64>* %A
+	%tmp2 = load <2 x i32>* %B
+	%tmp3 = zext <2 x i32> %tmp2 to <2 x i64>
+; CHECK: vaddw.u32	q8, q8, d18     @ encoding: [0xa2,0x01,0xe0,0xf3]
+	%tmp4 = add <2 x i64> %tmp1, %tmp3
+	ret <2 x i64> %tmp4
+}