Add a V8FP instruction 'vcvt{b,t}' to convert between half and double precision.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@185620 91177308-0d34-0410-b5e6-96231b3b80d8

6 files changed, 125 insertions, 1 deletions

diff --git a/lib/Target/ARM/ARMInstrInfo.td b/lib/Target/ARM/ARMInstrInfo.td
index 4d550ee..ed68b4e 100644
--- a/lib/Target/ARM/ARMInstrInfo.td
+++ b/lib/Target/ARM/ARMInstrInfo.td
@@ -203,6 +203,8 @@ def HasVFP3          : Predicate<"Subtarget->hasVFP3()">,
                                  AssemblerPredicate<"FeatureVFP3", "VFP3">;
 def HasVFP4          : Predicate<"Subtarget->hasVFP4()">,
                                  AssemblerPredicate<"FeatureVFP4", "VFP4">;
+def HasV8FP          : Predicate<"Subtarget->hasV8FP()">,
+                                 AssemblerPredicate<"FeatureV8FP", "V8FP">;
 def HasNEON          : Predicate<"Subtarget->hasNEON()">,
                                  AssemblerPredicate<"FeatureNEON", "NEON">;
 def HasFP16          : Predicate<"Subtarget->hasFP16()">,
diff --git a/lib/Target/ARM/ARMInstrVFP.td b/lib/Target/ARM/ARMInstrVFP.td
index 597b74a..e1d470f 100644
--- a/lib/Target/ARM/ARMInstrVFP.td
+++ b/lib/Target/ARM/ARMInstrVFP.td
@@ -468,7 +468,7 @@ def VCVTSD  : VFPAI<(outs SPR:$Sd), (ins DPR:$Dm), VFPUnaryFrm,
   let Inst{4}     = 0;
 }
 
-// Between half-precision and single-precision.  For disassembly only.
+// Between half, single and double-precision.  For disassembly only.
 
 // FIXME: Verify encoding after integrated assembler is working.
 def VCVTBHS: ASuI<0b11101, 0b11, 0b0010, 0b01, 0, (outs SPR:$Sd), (ins SPR:$Sm),
@@ -493,6 +493,60 @@ def VCVTTSH: ASuI<0b11101, 0b11, 0b0011, 0b11, 0, (outs SPR:$Sd), (ins SPR:$Sm),
                  /* FIXME */ IIC_fpCVTHS, "vcvtt", ".f16.f32\t$Sd, $Sm",
                  [/* For disassembly only; pattern left blank */]>;
 
+def VCVTBHD : ADuI<0b11101, 0b11, 0b0010, 0b01, 0,
+                   (outs DPR:$Dd), (ins SPR:$Sm),
+                   NoItinerary, "vcvtb", ".f64.f16\t$Dd, $Sm",
+                   []>, Requires<[HasV8FP]> {
+  // Instruction operands.
+  bits<5> Sm;
+
+  // Encode instruction operands.
+  let Inst{3-0} = Sm{4-1};
+  let Inst{5}   = Sm{0};
+}
+
+def VCVTBDH : ADuI<0b11101, 0b11, 0b0011, 0b01, 0,
+                   (outs SPR:$Sd), (ins DPR:$Dm),
+                   NoItinerary, "vcvtb", ".f16.f64\t$Sd, $Dm",
+                   []>, Requires<[HasV8FP]> {
+  // Instruction operands.
+  bits<5> Sd;
+  bits<5> Dm;
+
+  // Encode instruction operands.
+  let Inst{3-0}     = Dm{3-0};
+  let Inst{5}       = Dm{4};
+  let Inst{15-12}   = Sd{4-1};
+  let Inst{22}      = Sd{0};
+}
+
+def VCVTTHD : ADuI<0b11101, 0b11, 0b0010, 0b11, 0,
+                   (outs DPR:$Dd), (ins SPR:$Sm),
+                   NoItinerary, "vcvtt", ".f64.f16\t$Dd, $Sm",
+                   []>, Requires<[HasV8FP]> {
+  // Instruction operands.
+  bits<5> Sm;
+
+  // Encode instruction operands.
+  let Inst{3-0} = Sm{4-1};
+  let Inst{5}   = Sm{0};
+}
+
+def VCVTTDH : ADuI<0b11101, 0b11, 0b0011, 0b11, 0,
+                   (outs SPR:$Sd), (ins DPR:$Dm),
+                   NoItinerary, "vcvtt", ".f16.f64\t$Sd, $Dm",
+                   []>, Requires<[HasV8FP]> {
+  // Instruction operands.
+  bits<5> Sd;
+  bits<5> Dm;
+
+  // Encode instruction operands.
+  let Inst{15-12} = Sd{4-1};
+  let Inst{22}    = Sd{0};
+  let Inst{3-0}   = Dm{3-0};
+  let Inst{5}     = Dm{4};
+}
+
 def VNEGD  : ADuI<0b11101, 0b11, 0b0001, 0b01, 0,
                   (outs DPR:$Dd), (ins DPR:$Dm),
                   IIC_fpUNA64, "vneg", ".f64\t$Dd, $Dm",
diff --git a/test/MC/ARM/invalid-v8fp.s b/test/MC/ARM/invalid-v8fp.s
new file mode 100644
index 0000000..e2dbb8b
--- /dev/null
+++ b/test/MC/ARM/invalid-v8fp.s
@@ -0,0 +1,10 @@
+@ RUN: llvm-mc -triple armv7 -show-encoding < %s | FileCheck %s
+
+@ VCVT{B,T}
+
+  vcvtt.f64.f16 d3, s1
+@ CHECK-NOT: vcvtt.f64.f16 d3, s1      @ encoding: [0xe0,0x3b,0xb2,0xee]
+  vcvtt.f16.f64 s5, d12
+@ CHECK-NOT: vcvtt.f16.f64 s5, d12     @ encoding: [0xcc,0x2b,0xf3,0xee]
+
+
diff --git a/test/MC/ARM/v8fp.s b/test/MC/ARM/v8fp.s
new file mode 100644
index 0000000..f6bf7b9
--- /dev/null
+++ b/test/MC/ARM/v8fp.s
@@ -0,0 +1,23 @@
+@ RUN: llvm-mc -triple armv8 -mattr=+v8fp -show-encoding < %s | FileCheck %s
+
+@ VCVT{B,T}
+
+  vcvtt.f64.f16 d3, s1
+@ CHECK: vcvtt.f64.f16 d3, s1      @ encoding: [0xe0,0x3b,0xb2,0xee]
+  vcvtt.f16.f64 s5, d12
+@ CHECK: vcvtt.f16.f64 s5, d12     @ encoding: [0xcc,0x2b,0xf3,0xee]
+
+  vcvtb.f64.f16 d3, s1
+@ CHECK: vcvtb.f64.f16 d3, s1     @ encoding: [0x60,0x3b,0xb2,0xee]
+  vcvtb.f16.f64 s4, d1
+@ CHECK: vcvtb.f16.f64 s4, d1     @ encoding: [0x41,0x2b,0xb3,0xee]
+
+  vcvttge.f64.f16 d3, s1
+@ CHECK: vcvttge.f64.f16 d3, s1      @ encoding: [0xe0,0x3b,0xb2,0xae]
+  vcvttgt.f16.f64 s5, d12
+@ CHECK: vcvttgt.f16.f64 s5, d12     @ encoding: [0xcc,0x2b,0xf3,0xce]
+
+  vcvtbeq.f64.f16 d3, s1
+@ CHECK: vcvtbeq.f64.f16 d3, s1     @ encoding: [0x60,0x3b,0xb2,0x0e]
+  vcvtblt.f16.f64 s4, d1
+@ CHECK: vcvtblt.f16.f64 s4, d1     @ encoding: [0x41,0x2b,0xb3,0xbe]
diff --git a/test/MC/Disassembler/ARM/invalid-v8fp.txt b/test/MC/Disassembler/ARM/invalid-v8fp.txt
new file mode 100644
index 0000000..dba48d8
--- /dev/null
+++ b/test/MC/Disassembler/ARM/invalid-v8fp.txt
@@ -0,0 +1,10 @@
+# RUN: llvm-mc -disassemble -triple armv7 -show-encoding < %s | FileCheck %s
+
+0xe0 0x3b 0xb2 0xee
+# CHECK-NOT: vcvtt.f64.f16 d3, s1
+
+0x41 0x2b 0xb3 0xee 
+# CHECK-NOT: vcvtb.f16.f64 s4, d1
+
+0x41 0x2b 0xb3 0xbe
+# CHECK-NOT: vcvtblt.f16.f64 s4, d1 
diff --git a/test/MC/Disassembler/ARM/v8fp.txt b/test/MC/Disassembler/ARM/v8fp.txt
new file mode 100644
index 0000000..76125da
--- /dev/null
+++ b/test/MC/Disassembler/ARM/v8fp.txt
@@ -0,0 +1,25 @@
+# RUN: llvm-mc -disassemble -triple armv8 -mattr=+v8fp -show-encoding < %s | FileCheck %s
+
+0xe0 0x3b 0xb2 0xee
+# CHECK: vcvtt.f64.f16 d3, s1
+
+0xcc 0x2b 0xf3 0xee
+# CHECK: vcvtt.f16.f64 s5, d12
+
+0x60 0x3b 0xb2 0xee
+# CHECK: vcvtb.f64.f16 d3, s1
+
+0x41 0x2b 0xb3 0xee
+# CHECK: vcvtb.f16.f64 s4, d1
+
+0xe0 0x3b 0xb2 0xae
+# CHECK: vcvttge.f64.f16 d3, s1
+
+0xcc 0x2b 0xf3 0xce
+# CHECK: vcvttgt.f16.f64 s5, d12
+
+0x60 0x3b 0xb2 0x0e
+# CHECK: vcvtbeq.f64.f16 d3, s1
+
+0x41 0x2b 0xb3 0xbe
+# CHECK: vcvtblt.f16.f64 s4, d1