From d17728f4fd670253cafdbcca6b5c5ebd08ea8bfd Mon Sep 17 00:00:00 2001
From: Pavel Iliin <Pavel.Iliin@arm.com>
Date: Thu, 23 Apr 2020 22:03:42 +0100
Subject: [PATCH] [AArch64][FIX] FPR16_lo for f16 indexed patterns.

---
 lib/Target/AArch64/AArch64InstrFormats.td      |  8 ++++----
 lib/Target/AArch64/AArch64RegisterBankInfo.cpp |  2 ++
 lib/Target/AArch64/AArch64RegisterInfo.cpp     |  1 +
 lib/Target/AArch64/AArch64RegisterInfo.td      | 11 +++++++++--
 4 files changed, 16 insertions(+), 6 deletions(-)

diff --git a/lib/Target/AArch64/AArch64InstrFormats.td b/lib/Target/AArch64/AArch64InstrFormats.td
index 13f086ff1f5..1743b0c68eb 100644
--- a/lib/Target/AArch64/AArch64InstrFormats.td
+++ b/lib/Target/AArch64/AArch64InstrFormats.td
@@ -8073,9 +8073,9 @@ multiclass SIMDFPIndexedTiedPatterns<string INST, SDPatternOperator OpNode> {
             (!cast<Instruction>(INST # "v8i16_indexed")
                 V128:$Rd, V128:$Rn, V128_lo:$Rm, VectorIndexH:$idx)>;
   def : Pat<(v8f16 (OpNode (v8f16 V128:$Rd), (v8f16 V128:$Rn),
-                           (AArch64dup (f16 FPR16Op:$Rm)))),
+                           (AArch64dup (f16 FPR16Op_lo:$Rm)))),
             (!cast<Instruction>(INST # "v8i16_indexed") V128:$Rd, V128:$Rn,
-                (SUBREG_TO_REG (i32 0), FPR16Op:$Rm, hsub), (i64 0))>;
+                (SUBREG_TO_REG (i32 0), FPR16Op_lo:$Rm, hsub), (i64 0))>;
 
   def : Pat<(v4f16 (OpNode (v4f16 V64:$Rd), (v4f16 V64:$Rn),
                            (AArch64duplane16 (v8f16 V128_lo:$Rm),
@@ -8083,9 +8083,9 @@ multiclass SIMDFPIndexedTiedPatterns<string INST, SDPatternOperator OpNode> {
             (!cast<Instruction>(INST # "v4i16_indexed")
                 V64:$Rd, V64:$Rn, V128_lo:$Rm, VectorIndexH:$idx)>;
   def : Pat<(v4f16 (OpNode (v4f16 V64:$Rd), (v4f16 V64:$Rn),
-                           (AArch64dup (f16 FPR16Op:$Rm)))),
+                           (AArch64dup (f16 FPR16Op_lo:$Rm)))),
             (!cast<Instruction>(INST # "v4i16_indexed") V64:$Rd, V64:$Rn,
-                (SUBREG_TO_REG (i32 0), FPR16Op:$Rm, hsub), (i64 0))>;
+                (SUBREG_TO_REG (i32 0), FPR16Op_lo:$Rm, hsub), (i64 0))>;
 
   def : Pat<(f16 (OpNode (f16 FPR16:$Rd), (f16 FPR16:$Rn),
                          (vector_extract (v8f16 V128_lo:$Rm), VectorIndexH:$idx))),
diff --git a/lib/Target/AArch64/AArch64RegisterBankInfo.cpp b/lib/Target/AArch64/AArch64RegisterBankInfo.cpp
index e671ff21621..77998650043 100644
--- a/lib/Target/AArch64/AArch64RegisterBankInfo.cpp
+++ b/lib/Target/AArch64/AArch64RegisterBankInfo.cpp
@@ -231,6 +231,8 @@ AArch64RegisterBankInfo::getRegBankFromRegClass(const TargetRegisterClass &RC,
   switch (RC.getID()) {
   case AArch64::FPR8RegClassID:
   case AArch64::FPR16RegClassID:
+  case AArch64::FPR16_loRegClassID:
+  case AArch64::FPR32_with_hsub_in_FPR16_loRegClassID:
   case AArch64::FPR32RegClassID:
   case AArch64::FPR64RegClassID:
   case AArch64::FPR64_loRegClassID:
diff --git a/lib/Target/AArch64/AArch64RegisterInfo.cpp b/lib/Target/AArch64/AArch64RegisterInfo.cpp
index c8556c21195..6c49542e4ff 100644
--- a/lib/Target/AArch64/AArch64RegisterInfo.cpp
+++ b/lib/Target/AArch64/AArch64RegisterInfo.cpp
@@ -597,6 +597,7 @@ unsigned AArch64RegisterInfo::getRegPressureLimit(const TargetRegisterClass *RC,
 
   case AArch64::FPR128_loRegClassID:
   case AArch64::FPR64_loRegClassID:
+  case AArch64::FPR16_loRegClassID:
     return 16;
   }
 }
diff --git a/lib/Target/AArch64/AArch64RegisterInfo.td b/lib/Target/AArch64/AArch64RegisterInfo.td
index 4cd01234a6b..93b6aa0cdb7 100644
--- a/lib/Target/AArch64/AArch64RegisterInfo.td
+++ b/lib/Target/AArch64/AArch64RegisterInfo.td
@@ -425,6 +425,9 @@ def FPR8  : RegisterClass<"AArch64", [untyped], 8, (sequence "B%u", 0, 31)> {
 def FPR16 : RegisterClass<"AArch64", [f16], 16, (sequence "H%u", 0, 31)> {
   let Size = 16;
 }
+def FPR16_lo : RegisterClass<"AArch64", [f16], 16, (trunc FPR16, 16)> {
+  let Size = 16;
+}
 def FPR32 : RegisterClass<"AArch64", [f32, i32], 32,(sequence "S%u", 0, 31)>;
 def FPR64 : RegisterClass<"AArch64", [f64, i64, v2f32, v1f64, v8i8, v4i16, v2i32,
                                     v1i64, v4f16],
@@ -648,6 +651,10 @@ def FPR16Op  : RegisterOperand<FPR16, "printOperand"> {
   let ParserMatchClass = FPRAsmOperand<"FPR16">;
 }
 
+def FPR16Op_lo  : RegisterOperand<FPR16_lo, "printOperand"> {
+  let ParserMatchClass = FPRAsmOperand<"FPR16_lo">;
+}
+
 def FPR32Op  : RegisterOperand<FPR32, "printOperand"> {
   let ParserMatchClass = FPRAsmOperand<"FPR32">;
 }
@@ -671,11 +678,11 @@ def XSeqPairs : RegisterTuples<[sube64, subo64],
                                [(decimate (rotl GPR64, 0), 2),
                                 (decimate (rotl GPR64, 1), 2)]>;
 
-def WSeqPairsClass   : RegisterClass<"AArch64", [untyped], 32, 
+def WSeqPairsClass   : RegisterClass<"AArch64", [untyped], 32,
                                      (add WSeqPairs)>{
   let Size = 64;
 }
-def XSeqPairsClass   : RegisterClass<"AArch64", [untyped], 64, 
+def XSeqPairsClass   : RegisterClass<"AArch64", [untyped], 64,
                                      (add XSeqPairs)>{
   let Size = 128;
 }