[ARM] Remove hasSideEffects from FP converts

Whether an instruction is deemed to have side effects in determined by
whether it has a tblgen pattern that emits a single instruction.
Because of the way a lot of the the vcvt instructions are specified
either in dagtodag code or with patterns that emit multiple
instructions, they don't get marked as not having side effects.

This just marks them as not having side effects manually. It can help
especially with instruction scheduling, to not create artificial
barriers, but one of these tests also managed to produce fewer
instructions.

Differential Revision: https://reviews.llvm.org/D81639

diff --git a/llvm/lib/Target/ARM/ARMInstrVFP.td b/llvm/lib/Target/ARM/ARMInstrVFP.td
index bc6f852..8a652c1 100644 (file)
--- a/llvm/lib/Target/ARM/ARMInstrVFP.td
+++ b/llvm/lib/Target/ARM/ARMInstrVFP.td
@@ -708,6 +708,7 @@ def VCVTDS  : ASuI<0b11101, 0b11, 0b0111, 0b11, 0,
   let Inst{22}    = Dd{4};
 
   let Predicates = [HasVFP2, HasDPVFP];
+  let hasSideEffects = 0;
 }
 
 // Special case encoding: bits 11-8 is 0b1011.
@@ -732,9 +733,11 @@ def VCVTSD  : VFPAI<(outs SPR:$Sd), (ins DPR:$Dm), VFPUnaryFrm,
   let Inst{4}     = 0;
 
   let Predicates = [HasVFP2, HasDPVFP];
+  let hasSideEffects = 0;
 }
 
 // Between half, single and double-precision.
+let hasSideEffects = 0 in
 def VCVTBHS: ASuI<0b11101, 0b11, 0b0010, 0b01, 0, (outs SPR:$Sd), (ins SPR:$Sm),
                  /* FIXME */ IIC_fpCVTSH, "vcvtb", ".f32.f16\t$Sd, $Sm",
                  [/* Intentionally left blank, see patterns below */]>,
@@ -746,6 +749,7 @@ def : FP16Pat<(f32 (fpextend (f16 HPR:$Sm))),
 def : FP16Pat<(f16_to_fp GPR:$a),
               (VCVTBHS (COPY_TO_REGCLASS GPR:$a, SPR))>;
 
+let hasSideEffects = 0 in
 def VCVTBSH: ASuI<0b11101, 0b11, 0b0011, 0b01, 0, (outs SPR:$Sd), (ins SPR:$Sm),
                  /* FIXME */ IIC_fpCVTHS, "vcvtb", ".f16.f32\t$Sd, $Sm",
                  [/* Intentionally left blank, see patterns below */]>,
@@ -763,6 +767,7 @@ def : FP16Pat<(insertelt (v4f16 DPR:$src1), (f16 (fpround (f32 SPR:$src2))), imm
               (v4f16 (INSERT_SUBREG (v4f16 DPR:$src1), (VCVTBSH SPR:$src2),
                                     (SSubReg_f16_reg imm:$lane)))>;
 
+let hasSideEffects = 0 in
 def VCVTTHS: ASuI<0b11101, 0b11, 0b0010, 0b11, 0, (outs SPR:$Sd), (ins SPR:$Sm),
                  /* FIXME */ IIC_fpCVTSH, "vcvtt", ".f32.f16\t$Sd, $Sm",
                  [/* Intentionally left blank, see patterns below */]>,
@@ -776,6 +781,7 @@ def : FP16Pat<(f32 (fpextend (extractelt (v4f16 DPR:$src), imm_odd:$lane))),
                 (v2f32 (COPY_TO_REGCLASS (v4f16 DPR:$src), DPR_VFP2)),
                 (SSubReg_f16_reg imm_odd:$lane)))>;
 
+let hasSideEffects = 0 in
 def VCVTTSH: ASuI<0b11101, 0b11, 0b0011, 0b11, 0, (outs SPR:$Sd), (ins SPR:$Sm),
                  /* FIXME */ IIC_fpCVTHS, "vcvtt", ".f16.f32\t$Sd, $Sm",
                  [/* Intentionally left blank, see patterns below */]>,
@@ -801,6 +807,8 @@ def VCVTBHD : ADuI<0b11101, 0b11, 0b0010, 0b01, 0,
   // Encode instruction operands.
   let Inst{3-0} = Sm{4-1};
   let Inst{5}   = Sm{0};
+
+  let hasSideEffects = 0;
 }
 
 def : FullFP16Pat<(f64 (fpextend (f16 HPR:$Sm))),
@@ -824,6 +832,8 @@ def VCVTBDH : ADuI<0b11101, 0b11, 0b0011, 0b01, 0,
   let Inst{5}       = Dm{4};
   let Inst{15-12}   = Sd{4-1};
   let Inst{22}      = Sd{0};
+
+  let hasSideEffects = 0;
 }
 
 def : FullFP16Pat<(f16 (fpround DPR:$Dm)),
@@ -843,6 +853,8 @@ def VCVTTHD : ADuI<0b11101, 0b11, 0b0010, 0b11, 0,
   // Encode instruction operands.
   let Inst{3-0} = Sm{4-1};
   let Inst{5}   = Sm{0};
+
+  let hasSideEffects = 0;
 }
 
 def VCVTTDH : ADuI<0b11101, 0b11, 0b0011, 0b11, 0,
@@ -858,11 +870,13 @@ def VCVTTDH : ADuI<0b11101, 0b11, 0b0011, 0b11, 0,
   let Inst{22}    = Sd{0};
   let Inst{3-0}   = Dm{3-0};
   let Inst{5}     = Dm{4};
+
+  let hasSideEffects = 0;
 }
 
 multiclass vcvt_inst<string opc, bits<2> rm,
                      SDPatternOperator node = null_frag> {
-  let PostEncoderMethod = "", DecoderNamespace = "VFPV8" in {
+  let PostEncoderMethod = "", DecoderNamespace = "VFPV8", hasSideEffects = 0 in {
     def SH : AHuInp<0b11101, 0b11, 0b1100, 0b11, 0,
                     (outs SPR:$Sd), (ins HPR:$Sm),
                     NoItinerary, !strconcat("vcvt", opc, ".s32.f16\t$Sd, $Sm"),
@@ -1367,6 +1381,7 @@ class AVConv1IDs_Encode<bits<5> opcod1, bits<2> opcod2, bits<4> opcod3,
   let Inst{22}    = Dd{4};
 
   let Predicates = [HasVFP2, HasDPVFP];
+  let hasSideEffects = 0;
 }
 
 class AVConv1InSs_Encode<bits<5> opcod1, bits<2> opcod2, bits<4> opcod3,
@@ -1383,6 +1398,8 @@ class AVConv1InSs_Encode<bits<5> opcod1, bits<2> opcod2, bits<4> opcod3,
   let Inst{5}     = Sm{0};
   let Inst{15-12} = Sd{4-1};
   let Inst{22}    = Sd{0};
+
+  let hasSideEffects = 0;
 }
 
 class AVConv1IHs_Encode<bits<5> opcod1, bits<2> opcod2, bits<4> opcod3,
@@ -1402,6 +1419,7 @@ class AVConv1IHs_Encode<bits<5> opcod1, bits<2> opcod2, bits<4> opcod3,
   let Inst{22}    = Sd{0};
 
   let Predicates = [HasFullFP16];
+  let hasSideEffects = 0;
 }
 
 def VSITOD : AVConv1IDs_Encode<0b11101, 0b11, 0b1000, 0b1011,
@@ -1515,6 +1533,7 @@ class AVConv1IsD_Encode<bits<5> opcod1, bits<2> opcod2, bits<4> opcod3,
   let Inst{22}    = Sd{0};
 
   let Predicates = [HasVFP2, HasDPVFP];
+  let hasSideEffects = 0;
 }
 
 class AVConv1InsS_Encode<bits<5> opcod1, bits<2> opcod2, bits<4> opcod3,
@@ -1551,6 +1570,7 @@ class AVConv1IsH_Encode<bits<5> opcod1, bits<2> opcod2, bits<4> opcod3,
   let Inst{22}    = Sd{0};
 
   let Predicates = [HasFullFP16];
+  let hasSideEffects = 0;
 }
 
 // Always set Z bit in the instruction, i.e. "round towards zero" variants.
@@ -1730,6 +1750,8 @@ class AVConv1XInsS_Encode<bits<5> op1, bits<2> op2, bits<4> op3, bits<4> op4,
   // if dp_operation then UInt(D:Vd) else UInt(Vd:D);
   let Inst{22} = dst{0};
   let Inst{15-12} = dst{4-1};
+
+  let hasSideEffects = 0;
 }
 
 // Double Precision register
@@ -1742,6 +1764,7 @@ class AVConv1XInsD_Encode<bits<5> op1, bits<2> op2, bits<4> op3, bits<4> op4,
   let Inst{22} = dst{4};
   let Inst{15-12} = dst{3-0};
 
+  let hasSideEffects = 0;
   let Predicates = [HasVFP2, HasDPVFP];
 }
 
@@ -1940,7 +1963,9 @@ class BF16_VCVT<string opc, bits<2> op7_6>
   let Inst{11-8}  = 0b1001;
   let Inst{7-6}   = op7_6;
   let Inst{4}     = 0;
+
   let DecoderNamespace = "VFPV8";
+  let hasSideEffects = 0;
 }
 
 def BF16_VCVTB : BF16_VCVT<"vcvtb", 0b01>;

diff --git a/llvm/test/CodeGen/ARM/cmov_fp16.ll b/llvm/test/CodeGen/ARM/cmov_fp16.ll
index 2c368d1..805955d 100644 (file)
--- a/llvm/test/CodeGen/ARM/cmov_fp16.ll
+++ b/llvm/test/CodeGen/ARM/cmov_fp16.ll
@@ -1,14 +1,15 @@
-; RUN: llc -mtriple=thumbv8.1m.main-none-none-eabi -mattr=+fullfp16 %s -o - | FileCheck %s --check-prefixes CHECK-THUMB,CHECK
-; RUN: llc -mtriple=armv8.2a-arm-none-eabi -mattr=+fullfp16 %s -o - | FileCheck %s --check-prefixes CHECK-ARM,CHECK
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=thumbv8.1m.main-none-none-eabi -mattr=+fullfp16 %s -o - | FileCheck %s --check-prefixes CHECK,CHECK-THUMB
+; RUN: llc -mtriple=armv8.2a-arm-none-eabi -mattr=+fullfp16 %s -o - | FileCheck %s --check-prefixes CHECK,CHECK-ARM
 
 define i32 @test_ne(i32 %x, i32 %y, i32 %a, i32 %b) {
 ; CHECK-LABEL: test_ne:
 ; CHECK:       @ %bb.0: @ %entry
-; CHECK-NEXT:    vmov s2, r0
-; CHECK-NEXT:    cmp r2, r3
 ; CHECK-NEXT:    vmov s0, r1
-; CHECK-NEXT:    vcvt.f16.u32 s2, s2
+; CHECK-NEXT:    cmp r2, r3
+; CHECK-NEXT:    vmov s2, r0
 ; CHECK-NEXT:    vcvt.f16.u32 s0, s0
+; CHECK-NEXT:    vcvt.f16.u32 s2, s2
 ; CHECK-NEXT:    vseleq.f16 s0, s0, s2
 ; CHECK-NEXT:    vmov.f16 r0, s0
 ; CHECK-NEXT:    bx lr
@@ -25,11 +26,11 @@ entry:
 define i32 @test_eq(i32 %x, i32 %y, i32 %a, i32 %b) {
 ; CHECK-LABEL: test_eq:
 ; CHECK:       @ %bb.0: @ %entry
-; CHECK-NEXT:    vmov s2, r1
-; CHECK-NEXT:    cmp r2, r3
 ; CHECK-NEXT:    vmov s0, r0
-; CHECK-NEXT:    vcvt.f16.u32 s2, s2
+; CHECK-NEXT:    cmp r2, r3
+; CHECK-NEXT:    vmov s2, r1
 ; CHECK-NEXT:    vcvt.f16.u32 s0, s0
+; CHECK-NEXT:    vcvt.f16.u32 s2, s2
 ; CHECK-NEXT:    vseleq.f16 s0, s0, s2
 ; CHECK-NEXT:    vmov.f16 r0, s0
 ; CHECK-NEXT:    bx lr
@@ -46,11 +47,11 @@ entry:
 define i32 @test_gt(i32 %x, i32 %y, i32 %a, i32 %b) {
 ; CHECK-LABEL: test_gt:
 ; CHECK:       @ %bb.0: @ %entry
-; CHECK-NEXT:    vmov s2, r1
-; CHECK-NEXT:    cmp r2, r3
 ; CHECK-NEXT:    vmov s0, r0
-; CHECK-NEXT:    vcvt.f16.u32 s2, s2
+; CHECK-NEXT:    cmp r2, r3
+; CHECK-NEXT:    vmov s2, r1
 ; CHECK-NEXT:    vcvt.f16.u32 s0, s0
+; CHECK-NEXT:    vcvt.f16.u32 s2, s2
 ; CHECK-NEXT:    vselgt.f16 s0, s0, s2
 ; CHECK-NEXT:    vmov.f16 r0, s0
 ; CHECK-NEXT:    bx lr
@@ -67,11 +68,11 @@ entry:
 define i32 @test_ge(i32 %x, i32 %y, i32 %a, i32 %b) {
 ; CHECK-LABEL: test_ge:
 ; CHECK:       @ %bb.0: @ %entry
-; CHECK-NEXT:    vmov s2, r1
-; CHECK-NEXT:    cmp r2, r3
 ; CHECK-NEXT:    vmov s0, r0
-; CHECK-NEXT:    vcvt.f16.u32 s2, s2
+; CHECK-NEXT:    cmp r2, r3
+; CHECK-NEXT:    vmov s2, r1
 ; CHECK-NEXT:    vcvt.f16.u32 s0, s0
+; CHECK-NEXT:    vcvt.f16.u32 s2, s2
 ; CHECK-NEXT:    vselge.f16 s0, s0, s2
 ; CHECK-NEXT:    vmov.f16 r0, s0
 ; CHECK-NEXT:    bx lr
@@ -88,11 +89,11 @@ entry:
 define i32 @test_lt(i32 %x, i32 %y, i32 %a, i32 %b) {
 ; CHECK-LABEL: test_lt:
 ; CHECK:       @ %bb.0: @ %entry
-; CHECK-NEXT:    vmov s2, r0
-; CHECK-NEXT:    cmp r2, r3
 ; CHECK-NEXT:    vmov s0, r1
-; CHECK-NEXT:    vcvt.f16.u32 s2, s2
+; CHECK-NEXT:    cmp r2, r3
+; CHECK-NEXT:    vmov s2, r0
 ; CHECK-NEXT:    vcvt.f16.u32 s0, s0
+; CHECK-NEXT:    vcvt.f16.u32 s2, s2
 ; CHECK-NEXT:    vselge.f16 s0, s0, s2
 ; CHECK-NEXT:    vmov.f16 r0, s0
 ; CHECK-NEXT:    bx lr
@@ -109,11 +110,11 @@ entry:
 define i32 @test_le(i32 %x, i32 %y, i32 %a, i32 %b) {
 ; CHECK-LABEL: test_le:
 ; CHECK:       @ %bb.0: @ %entry
-; CHECK-NEXT:    vmov s2, r0
-; CHECK-NEXT:    cmp r2, r3
 ; CHECK-NEXT:    vmov s0, r1
-; CHECK-NEXT:    vcvt.f16.u32 s2, s2
+; CHECK-NEXT:    cmp r2, r3
+; CHECK-NEXT:    vmov s2, r0
 ; CHECK-NEXT:    vcvt.f16.u32 s0, s0
+; CHECK-NEXT:    vcvt.f16.u32 s2, s2
 ; CHECK-NEXT:    vselgt.f16 s0, s0, s2
 ; CHECK-NEXT:    vmov.f16 r0, s0
 ; CHECK-NEXT:    bx lr
@@ -130,11 +131,11 @@ entry:
 define i32 @test_hi(i32 %x, i32 %y, i32 %a, i32 %b) {
 ; CHECK-THUMB-LABEL: test_hi:
 ; CHECK-THUMB:       @ %bb.0: @ %entry
-; CHECK-THUMB-NEXT:    vmov s2, r0
-; CHECK-THUMB-NEXT:    cmp r2, r3
 ; CHECK-THUMB-NEXT:    vmov s0, r1
-; CHECK-THUMB-NEXT:    vcvt.f16.u32 s2, s2
+; CHECK-THUMB-NEXT:    cmp r2, r3
+; CHECK-THUMB-NEXT:    vmov s2, r0
 ; CHECK-THUMB-NEXT:    vcvt.f16.u32 s0, s0
+; CHECK-THUMB-NEXT:    vcvt.f16.u32 s2, s2
 ; CHECK-THUMB-NEXT:    it hi
 ; CHECK-THUMB-NEXT:    vmovhi.f32 s0, s2
 ; CHECK-THUMB-NEXT:    vmov.f16 r0, s0
@@ -142,11 +143,11 @@ define i32 @test_hi(i32 %x, i32 %y, i32 %a, i32 %b) {
 ;
 ; CHECK-ARM-LABEL: test_hi:
 ; CHECK-ARM:       @ %bb.0: @ %entry
-; CHECK-ARM-NEXT:    vmov s2, r0
-; CHECK-ARM-NEXT:    cmp r2, r3
 ; CHECK-ARM-NEXT:    vmov s0, r1
-; CHECK-ARM-NEXT:    vcvt.f16.u32 s2, s2
+; CHECK-ARM-NEXT:    cmp r2, r3
+; CHECK-ARM-NEXT:    vmov s2, r0
 ; CHECK-ARM-NEXT:    vcvt.f16.u32 s0, s0
+; CHECK-ARM-NEXT:    vcvt.f16.u32 s2, s2
 ; CHECK-ARM-NEXT:    vmovhi.f32 s0, s2
 ; CHECK-ARM-NEXT:    vmov.f16 r0, s0
 ; CHECK-ARM-NEXT:    bx lr
@@ -163,11 +164,11 @@ entry:
 define i32 @test_hs(i32 %x, i32 %y, i32 %a, i32 %b) {
 ; CHECK-THUMB-LABEL: test_hs:
 ; CHECK-THUMB:       @ %bb.0: @ %entry
-; CHECK-THUMB-NEXT:    vmov s2, r0
-; CHECK-THUMB-NEXT:    cmp r2, r3
 ; CHECK-THUMB-NEXT:    vmov s0, r1
-; CHECK-THUMB-NEXT:    vcvt.f16.u32 s2, s2
+; CHECK-THUMB-NEXT:    cmp r2, r3
+; CHECK-THUMB-NEXT:    vmov s2, r0
 ; CHECK-THUMB-NEXT:    vcvt.f16.u32 s0, s0
+; CHECK-THUMB-NEXT:    vcvt.f16.u32 s2, s2
 ; CHECK-THUMB-NEXT:    it hs
 ; CHECK-THUMB-NEXT:    vmovhs.f32 s0, s2
 ; CHECK-THUMB-NEXT:    vmov.f16 r0, s0
@@ -175,11 +176,11 @@ define i32 @test_hs(i32 %x, i32 %y, i32 %a, i32 %b) {
 ;
 ; CHECK-ARM-LABEL: test_hs:
 ; CHECK-ARM:       @ %bb.0: @ %entry
-; CHECK-ARM-NEXT:    vmov s2, r0
-; CHECK-ARM-NEXT:    cmp r2, r3
 ; CHECK-ARM-NEXT:    vmov s0, r1
-; CHECK-ARM-NEXT:    vcvt.f16.u32 s2, s2
+; CHECK-ARM-NEXT:    cmp r2, r3
+; CHECK-ARM-NEXT:    vmov s2, r0
 ; CHECK-ARM-NEXT:    vcvt.f16.u32 s0, s0
+; CHECK-ARM-NEXT:    vcvt.f16.u32 s2, s2
 ; CHECK-ARM-NEXT:    vmovhs.f32 s0, s2
 ; CHECK-ARM-NEXT:    vmov.f16 r0, s0
 ; CHECK-ARM-NEXT:    bx lr
@@ -196,11 +197,11 @@ entry:
 define i32 @test_lo(i32 %x, i32 %y, i32 %a, i32 %b) {
 ; CHECK-THUMB-LABEL: test_lo:
 ; CHECK-THUMB:       @ %bb.0: @ %entry
-; CHECK-THUMB-NEXT:    vmov s2, r0
-; CHECK-THUMB-NEXT:    cmp r2, r3
 ; CHECK-THUMB-NEXT:    vmov s0, r1
-; CHECK-THUMB-NEXT:    vcvt.f16.u32 s2, s2
+; CHECK-THUMB-NEXT:    cmp r2, r3
+; CHECK-THUMB-NEXT:    vmov s2, r0
 ; CHECK-THUMB-NEXT:    vcvt.f16.u32 s0, s0
+; CHECK-THUMB-NEXT:    vcvt.f16.u32 s2, s2
 ; CHECK-THUMB-NEXT:    it lo
 ; CHECK-THUMB-NEXT:    vmovlo.f32 s0, s2
 ; CHECK-THUMB-NEXT:    vmov.f16 r0, s0
@@ -208,11 +209,11 @@ define i32 @test_lo(i32 %x, i32 %y, i32 %a, i32 %b) {
 ;
 ; CHECK-ARM-LABEL: test_lo:
 ; CHECK-ARM:       @ %bb.0: @ %entry
-; CHECK-ARM-NEXT:    vmov s2, r0
-; CHECK-ARM-NEXT:    cmp r2, r3
 ; CHECK-ARM-NEXT:    vmov s0, r1
-; CHECK-ARM-NEXT:    vcvt.f16.u32 s2, s2
+; CHECK-ARM-NEXT:    cmp r2, r3
+; CHECK-ARM-NEXT:    vmov s2, r0
 ; CHECK-ARM-NEXT:    vcvt.f16.u32 s0, s0
+; CHECK-ARM-NEXT:    vcvt.f16.u32 s2, s2
 ; CHECK-ARM-NEXT:    vmovlo.f32 s0, s2
 ; CHECK-ARM-NEXT:    vmov.f16 r0, s0
 ; CHECK-ARM-NEXT:    bx lr
@@ -229,11 +230,11 @@ entry:
 define i32 @test_ls(i32 %x, i32 %y, i32 %a, i32 %b) {
 ; CHECK-THUMB-LABEL: test_ls:
 ; CHECK-THUMB:       @ %bb.0: @ %entry
-; CHECK-THUMB-NEXT:    vmov s2, r0
-; CHECK-THUMB-NEXT:    cmp r2, r3
 ; CHECK-THUMB-NEXT:    vmov s0, r1
-; CHECK-THUMB-NEXT:    vcvt.f16.u32 s2, s2
+; CHECK-THUMB-NEXT:    cmp r2, r3
+; CHECK-THUMB-NEXT:    vmov s2, r0
 ; CHECK-THUMB-NEXT:    vcvt.f16.u32 s0, s0
+; CHECK-THUMB-NEXT:    vcvt.f16.u32 s2, s2
 ; CHECK-THUMB-NEXT:    it ls
 ; CHECK-THUMB-NEXT:    vmovls.f32 s0, s2
 ; CHECK-THUMB-NEXT:    vmov.f16 r0, s0
@@ -241,11 +242,11 @@ define i32 @test_ls(i32 %x, i32 %y, i32 %a, i32 %b) {
 ;
 ; CHECK-ARM-LABEL: test_ls:
 ; CHECK-ARM:       @ %bb.0: @ %entry
-; CHECK-ARM-NEXT:    vmov s2, r0
-; CHECK-ARM-NEXT:    cmp r2, r3
 ; CHECK-ARM-NEXT:    vmov s0, r1
-; CHECK-ARM-NEXT:    vcvt.f16.u32 s2, s2
+; CHECK-ARM-NEXT:    cmp r2, r3
+; CHECK-ARM-NEXT:    vmov s2, r0
 ; CHECK-ARM-NEXT:    vcvt.f16.u32 s0, s0
+; CHECK-ARM-NEXT:    vcvt.f16.u32 s2, s2
 ; CHECK-ARM-NEXT:    vmovls.f32 s0, s2
 ; CHECK-ARM-NEXT:    vmov.f16 r0, s0
 ; CHECK-ARM-NEXT:    bx lr

diff --git a/llvm/test/CodeGen/ARM/fp16-args.ll b/llvm/test/CodeGen/ARM/fp16-args.ll
index e858661..7ed1e88 100644 (file)
--- a/llvm/test/CodeGen/ARM/fp16-args.ll
+++ b/llvm/test/CodeGen/ARM/fp16-args.ll
@@ -11,10 +11,10 @@
 define half @foo(half %a, half %b) {
 ; SOFT-LABEL: foo:
 ; SOFT:       @ %bb.0: @ %entry
-; SOFT-NEXT:    vmov s2, r1
 ; SOFT-NEXT:    vmov s0, r0
-; SOFT-NEXT:    vcvtb.f32.f16 s2, s2
+; SOFT-NEXT:    vmov s2, r1
 ; SOFT-NEXT:    vcvtb.f32.f16 s0, s0
+; SOFT-NEXT:    vcvtb.f32.f16 s2, s2
 ; SOFT-NEXT:    vadd.f32 s0, s0, s2
 ; SOFT-NEXT:    vcvtb.f16.f32 s0, s0
 ; SOFT-NEXT:    vmov r0, s0

diff --git a/llvm/test/CodeGen/ARM/fp16-bitcast.ll b/llvm/test/CodeGen/ARM/fp16-bitcast.ll
index 4254f2e..d26c2d9 100644 (file)
--- a/llvm/test/CodeGen/ARM/fp16-bitcast.ll
+++ b/llvm/test/CodeGen/ARM/fp16-bitcast.ll
@@ -40,10 +40,10 @@ entry:
 define half @addf16(half %a, half %b) {
 ; CHECK-VFPV4-SOFT-LABEL: addf16:
 ; CHECK-VFPV4-SOFT:       @ %bb.0: @ %entry
-; CHECK-VFPV4-SOFT-NEXT:    vmov s2, r1
 ; CHECK-VFPV4-SOFT-NEXT:    vmov s0, r0
-; CHECK-VFPV4-SOFT-NEXT:    vcvtb.f32.f16 s2, s2
+; CHECK-VFPV4-SOFT-NEXT:    vmov s2, r1
 ; CHECK-VFPV4-SOFT-NEXT:    vcvtb.f32.f16 s0, s0
+; CHECK-VFPV4-SOFT-NEXT:    vcvtb.f32.f16 s2, s2
 ; CHECK-VFPV4-SOFT-NEXT:    vadd.f32 s0, s0, s2
 ; CHECK-VFPV4-SOFT-NEXT:    vcvtb.f16.f32 s0, s0
 ; CHECK-VFPV4-SOFT-NEXT:    vmov r0, s0

diff --git a/llvm/test/CodeGen/ARM/fp16-fullfp16.ll b/llvm/test/CodeGen/ARM/fp16-fullfp16.ll
index e912cee..86a8caa 100644 (file)
--- a/llvm/test/CodeGen/ARM/fp16-fullfp16.ll
+++ b/llvm/test/CodeGen/ARM/fp16-fullfp16.ll
@@ -61,8 +61,8 @@ define arm_aapcs_vfpcc void @test_frem(half* %p, half* %q) {
 ; CHECK-LABEL: test_frem:
 ; CHECK:         .save {r4, lr}
 ; CHECK-NEXT:    push {r4, lr}
-; CHECK-NEXT:    vldr.16 s2, [r1]
 ; CHECK-NEXT:    vldr.16 s0, [r0]
+; CHECK-NEXT:    vldr.16 s2, [r1]
 ; CHECK-NEXT:    mov r4, r0
 ; CHECK-NEXT:    vcvtb.f32.f16 s0, s0
 ; CHECK-NEXT:    vcvtb.f32.f16 s1, s2
@@ -285,8 +285,8 @@ define void @test_pow(half* %p, half* %q) {
 ; CHECK-LABEL: test_pow:
 ; CHECK:         .save {r4, lr}
 ; CHECK-NEXT:    push {r4, lr}
-; CHECK-NEXT:    vldr.16 s2, [r1]
 ; CHECK-NEXT:    vldr.16 s0, [r0]
+; CHECK-NEXT:    vldr.16 s2, [r1]
 ; CHECK-NEXT:    mov r4, r0
 ; CHECK-NEXT:    vcvtb.f32.f16 s0, s0
 ; CHECK-NEXT:    vcvtb.f32.f16 s1, s2

diff --git a/llvm/test/CodeGen/ARM/fp16-instructions.ll b/llvm/test/CodeGen/ARM/fp16-instructions.ll
index 260dd12..5e3e446 100644 (file)
--- a/llvm/test/CodeGen/ARM/fp16-instructions.ll
+++ b/llvm/test/CodeGen/ARM/fp16-instructions.ll
@@ -72,10 +72,10 @@ entry:
 ; CHECK-SOFTFP-VFP3:  vadd.f32
 ; CHECK-SOFTFP-VFP3:  bl  __aeabi_f2h
 
-; CHECK-SOFTFP-FP16:  vmov          [[S2:s[0-9]]], r1
-; CHECK-SOFTFP-FP16:  vmov          [[S0:s[0-9]]], r0
-; CHECK-SOFTFP-FP16:  vcvtb.f32.f16 [[S2]], [[S2]]
-; CHECK-SOFTFP-FP16:  vcvtb.f32.f16 [[S0]], [[S0]]
+; CHECK-SOFTFP-FP16-DAG:  vmov          [[S0:s[0-9]]], r0
+; CHECK-SOFTFP-FP16-DAG:  vmov          [[S2:s[0-9]]], r1
+; CHECK-SOFTFP-FP16-DAG:  vcvtb.f32.f16 [[S0]], [[S0]]
+; CHECK-SOFTFP-FP16-DAG:  vcvtb.f32.f16 [[S2]], [[S2]]
 ; CHECK-SOFTFP-FP16:  vadd.f32      [[S0]], [[S0]], [[S2]]
 ; CHECK-SOFTFP-FP16:  vcvtb.f16.f32 [[S0]], [[S0]]
 ; CHECK-SOFTFP-FP16:  vmov  r0, s0
@@ -355,10 +355,10 @@ entry:
 ; CHECK-SOFTFP-VFP3:  vdiv.f32
 ; CHECK-SOFTFP-VFP3:  bl  __aeabi_f2h
 
-; CHECK-SOFTFP-FP16:  vmov          [[S2:s[0-9]]], r1
-; CHECK-SOFTFP-FP16:  vmov          [[S0:s[0-9]]], r0
-; CHECK-SOFTFP-FP16:  vcvtb.f32.f16 [[S2]], [[S2]]
-; CHECK-SOFTFP-FP16:  vcvtb.f32.f16 [[S0]], [[S0]]
+; CHECK-SOFTFP-FP16-DAG:  vmov          [[S0:s[0-9]]], r0
+; CHECK-SOFTFP-FP16-DAG:  vmov          [[S2:s[0-9]]], r1
+; CHECK-SOFTFP-FP16-DAG:  vcvtb.f32.f16 [[S0]], [[S0]]
+; CHECK-SOFTFP-FP16-DAG:  vcvtb.f32.f16 [[S2]], [[S2]]
 ; CHECK-SOFTFP-FP16:  vdiv.f32      [[S0]], [[S0]], [[S2]]
 ; CHECK-SOFTFP-FP16:  vcvtb.f16.f32 [[S0]], [[S0]]
 ; CHECK-SOFTFP-FP16:  vmov  r0, s0
@@ -577,10 +577,10 @@ entry:
 ; CHECK-SOFTFP-VFP3:  vmul.f32
 ; CHECK-SOFTFP-VFP3:  bl  __aeabi_f2h
 
-; CHECK-SOFTFP-FP16:  vmov          [[S2:s[0-9]]], r1
-; CHECK-SOFTFP-FP16:  vmov          [[S0:s[0-9]]], r0
-; CHECK-SOFTFP-FP16:  vcvtb.f32.f16 [[S2]], [[S2]]
-; CHECK-SOFTFP-FP16:  vcvtb.f32.f16 [[S0]], [[S0]]
+; CHECK-SOFTFP-FP16-DAG:  vmov          [[S0:s[0-9]]], r0
+; CHECK-SOFTFP-FP16-DAG:  vmov          [[S2:s[0-9]]], r1
+; CHECK-SOFTFP-FP16-DAG:  vcvtb.f32.f16 [[S0]], [[S0]]
+; CHECK-SOFTFP-FP16-DAG:  vcvtb.f32.f16 [[S2]], [[S2]]
 ; CHECK-SOFTFP-FP16:  vmul.f32      [[S0]], [[S0]], [[S2]]
 ; CHECK-SOFTFP-FP16:  vcvtb.f16.f32 [[S0]], [[S0]]
 ; CHECK-SOFTFP-FP16:  vmov  r0, s0
@@ -973,10 +973,10 @@ entry:
 ; CHECK-SOFTFP-VFP3:  vsub.f32
 ; CHECK-SOFTFP-VFP3:  bl  __aeabi_f2h
 
-; CHECK-SOFTFP-FP16:  vmov          [[S2:s[0-9]]], r1
-; CHECK-SOFTFP-FP16:  vmov          [[S0:s[0-9]]], r0
-; CHECK-SOFTFP-FP16:  vcvtb.f32.f16 [[S2]], [[S2]]
-; CHECK-SOFTFP-FP16:  vcvtb.f32.f16 [[S0]], [[S0]]
+; CHECK-SOFTFP-FP16-DAG:  vmov          [[S0:s[0-9]]], r0
+; CHECK-SOFTFP-FP16-DAG:  vmov          [[S2:s[0-9]]], r1
+; CHECK-SOFTFP-FP16-DAG:  vcvtb.f32.f16 [[S0]], [[S0]]
+; CHECK-SOFTFP-FP16-DAG:  vcvtb.f32.f16 [[S2]], [[S2]]
 ; CHECK-SOFTFP-FP16:  vsub.f32      [[S0]], [[S0]], [[S2]]
 ; CHECK-SOFTFP-FP16:  vcvtb.f16.f32 [[S0]], [[S0]]
 ; CHECK-SOFTFP-FP16:  vmov  r0, s0

diff --git a/llvm/test/CodeGen/ARM/no-fpscr-liveness.ll b/llvm/test/CodeGen/ARM/no-fpscr-liveness.ll
index 6575ab6..4fdb961 100644 (file)
--- a/llvm/test/CodeGen/ARM/no-fpscr-liveness.ll
+++ b/llvm/test/CodeGen/ARM/no-fpscr-liveness.ll
@@ -12,7 +12,9 @@ target triple = "thumbv7s-apple-ios"
 ; CHECK-LABEL: eggs:
 ; CHECK: sub sp, #8
 ; VMRS instruction comes before any other instruction writing FPSCR:
-; CHECK-NEXT: vmrs r0, fpscr
+; CHECK-NOT: vcmp
+; CHECK: vmrs {{r[0-9]}}, fpscr
+; CHECK; vcmp
 ; ...
 ; CHECK: add sp, #8
 ; CHECK: bx lr

diff --git a/llvm/test/CodeGen/ARM/vcvt.ll b/llvm/test/CodeGen/ARM/vcvt.ll
index f16c8dc..8db0856 100644 (file)
--- a/llvm/test/CodeGen/ARM/vcvt.ll
+++ b/llvm/test/CodeGen/ARM/vcvt.ll
@@ -296,13 +296,13 @@ define <4 x i16> @fix_double_to_i16(<4 x double> %in) {
 ; CHECK-NEXT:    vcvt.s32.f64 s0, d18
 ; CHECK-NEXT:    vmov r0, s0
 ; CHECK-NEXT:    vadd.f64 d20, d16, d16
-; CHECK-NEXT:    vadd.f64 d19, d19, d19
 ; CHECK-NEXT:    vadd.f64 d16, d17, d17
 ; CHECK-NEXT:    vcvt.s32.f64 s2, d20
-; CHECK-NEXT:    vcvt.s32.f64 s4, d19
 ; CHECK-NEXT:    vcvt.s32.f64 s6, d16
 ; CHECK-NEXT:    vmov.32 d16[0], r0
 ; CHECK-NEXT:    vmov r0, s2
+; CHECK-NEXT:    vadd.f64 d19, d19, d19
+; CHECK-NEXT:    vcvt.s32.f64 s4, d19
 ; CHECK-NEXT:    vmov.32 d17[0], r0
 ; CHECK-NEXT:    vmov r0, s4
 ; CHECK-NEXT:    vmov.32 d16[1], r0
@@ -355,10 +355,8 @@ define i32 @multi_sint(double %c, i32* nocapture %p, i32* nocapture %q) {
 ; CHECK:       @ %bb.0:
 ; CHECK-NEXT:    vmov d16, r0, r1
 ; CHECK-NEXT:    vcvt.s32.f64 s0, d16
+; CHECK-NEXT:    vmov r0, s0
 ; CHECK-NEXT:    vstr s0, [r2]
-; CHECK-NEXT:    vcvt.s32.f64 s0, d16
-; CHECK-NEXT:    vcvt.s32.f64 s2, d16
-; CHECK-NEXT:    vmov r0, s2
 ; CHECK-NEXT:    vstr s0, [r3]
 ; CHECK-NEXT:    mov pc, lr
   %conv = fptosi double %c to i32
@@ -372,10 +370,8 @@ define i32 @multi_uint(double %c, i32* nocapture %p, i32* nocapture %q) {
 ; CHECK:       @ %bb.0:
 ; CHECK-NEXT:    vmov d16, r0, r1
 ; CHECK-NEXT:    vcvt.u32.f64 s0, d16
+; CHECK-NEXT:    vmov r0, s0
 ; CHECK-NEXT:    vstr s0, [r2]
-; CHECK-NEXT:    vcvt.u32.f64 s0, d16
-; CHECK-NEXT:    vcvt.u32.f64 s2, d16
-; CHECK-NEXT:    vmov r0, s2
 ; CHECK-NEXT:    vstr s0, [r3]
 ; CHECK-NEXT:    mov pc, lr
   %conv = fptoui double %c to i32

diff --git a/llvm/test/CodeGen/Thumb2/LowOverheadLoops/mve-float-loops.ll b/llvm/test/CodeGen/Thumb2/LowOverheadLoops/mve-float-loops.ll
index b64c237..e69610f 100644 (file)
--- a/llvm/test/CodeGen/Thumb2/LowOverheadLoops/mve-float-loops.ll
+++ b/llvm/test/CodeGen/Thumb2/LowOverheadLoops/mve-float-loops.ll
@@ -710,10 +710,10 @@ define arm_aapcs_vfpcc void @float_int_mul(float* nocapture readonly %a, i32* no
 ; CHECK-NEXT:    @ =>This Inner Loop Header: Depth=1
 ; CHECK-NEXT:    ldr r4, [r6], #4
 ; CHECK-NEXT:    add.w r12, r12, #1
-; CHECK-NEXT:    vmov s0, r4
-; CHECK-NEXT:    vcvt.f32.s32 s0, s0
 ; CHECK-NEXT:    vldr s2, [r5]
 ; CHECK-NEXT:    adds r5, #4
+; CHECK-NEXT:    vmov s0, r4
+; CHECK-NEXT:    vcvt.f32.s32 s0, s0
 ; CHECK-NEXT:    vmul.f32 s0, s2, s0
 ; CHECK-NEXT:    vstr s0, [r7]
 ; CHECK-NEXT:    adds r7, #4
@@ -739,19 +739,19 @@ define arm_aapcs_vfpcc void @float_int_mul(float* nocapture readonly %a, i32* no
 ; CHECK-NEXT:    vmul.f32 s0, s2, s0
 ; CHECK-NEXT:    vstr s0, [r6]
 ; CHECK-NEXT:    vldr s0, [r1, #-4]
-; CHECK-NEXT:    vcvt.f32.s32 s0, s0
 ; CHECK-NEXT:    vldr s2, [r7, #4]
+; CHECK-NEXT:    vcvt.f32.s32 s0, s0
 ; CHECK-NEXT:    vmul.f32 s0, s2, s0
 ; CHECK-NEXT:    vstr s0, [r6, #4]
 ; CHECK-NEXT:    vldr s0, [r1]
-; CHECK-NEXT:    vcvt.f32.s32 s0, s0
 ; CHECK-NEXT:    vldr s2, [r7, #8]
+; CHECK-NEXT:    vcvt.f32.s32 s0, s0
 ; CHECK-NEXT:    vmul.f32 s0, s2, s0
 ; CHECK-NEXT:    vstr s0, [r6, #8]
 ; CHECK-NEXT:    vldr s0, [r1, #4]
 ; CHECK-NEXT:    add.w r1, r1, #16
-; CHECK-NEXT:    vcvt.f32.s32 s0, s0
 ; CHECK-NEXT:    vldr s2, [r7, #12]
+; CHECK-NEXT:    vcvt.f32.s32 s0, s0
 ; CHECK-NEXT:    vmul.f32 s0, s2, s0
 ; CHECK-NEXT:    vstr s0, [r6, #12]
 ; CHECK-NEXT:    bne .LBB3_12
@@ -1490,18 +1490,18 @@ define arm_aapcs_vfpcc float @half_half_mac(half* nocapture readonly %a, half* n
 ; CHECK-NEXT:    vmul.f16 s2, s4, s2
 ; CHECK-NEXT:    vldr.16 s4, [r2, #4]
 ; CHECK-NEXT:    vldr.16 s10, [r4]
-; CHECK-NEXT:    adds r3, #8
+; CHECK-NEXT:    vcvtb.f32.f16 s2, s2
 ; CHECK-NEXT:    vmul.f16 s4, s6, s4
 ; CHECK-NEXT:    vldr.16 s6, [r2, #2]
-; CHECK-NEXT:    add.w r12, r12, #4
+; CHECK-NEXT:    vcvtb.f32.f16 s4, s4
+; CHECK-NEXT:    adds r3, #8
 ; CHECK-NEXT:    vmul.f16 s6, s8, s6
 ; CHECK-NEXT:    vldr.16 s8, [r2]
+; CHECK-NEXT:    vcvtb.f32.f16 s6, s6
+; CHECK-NEXT:    add.w r12, r12, #4
 ; CHECK-NEXT:    vmul.f16 s8, s10, s8
 ; CHECK-NEXT:    vcvtb.f32.f16 s8, s8
-; CHECK-NEXT:    vcvtb.f32.f16 s6, s6
 ; CHECK-NEXT:    vadd.f32 s0, s0, s8
-; CHECK-NEXT:    vcvtb.f32.f16 s4, s4
-; CHECK-NEXT:    vcvtb.f32.f16 s2, s2
 ; CHECK-NEXT:    vadd.f32 s0, s0, s6
 ; CHECK-NEXT:    vadd.f32 s0, s0, s4
 ; CHECK-NEXT:    vadd.f32 s0, s0, s2
@@ -1647,18 +1647,18 @@ define arm_aapcs_vfpcc float @half_half_acc(half* nocapture readonly %a, half* n
 ; CHECK-NEXT:    vadd.f16 s2, s4, s2
 ; CHECK-NEXT:    vldr.16 s4, [r2, #4]
 ; CHECK-NEXT:    vldr.16 s10, [r4]
-; CHECK-NEXT:    adds r3, #8
+; CHECK-NEXT:    vcvtb.f32.f16 s2, s2
 ; CHECK-NEXT:    vadd.f16 s4, s6, s4
 ; CHECK-NEXT:    vldr.16 s6, [r2, #2]
-; CHECK-NEXT:    add.w r12, r12, #4
+; CHECK-NEXT:    vcvtb.f32.f16 s4, s4
+; CHECK-NEXT:    adds r3, #8
 ; CHECK-NEXT:    vadd.f16 s6, s8, s6
 ; CHECK-NEXT:    vldr.16 s8, [r2]
+; CHECK-NEXT:    vcvtb.f32.f16 s6, s6
+; CHECK-NEXT:    add.w r12, r12, #4
 ; CHECK-NEXT:    vadd.f16 s8, s10, s8
 ; CHECK-NEXT:    vcvtb.f32.f16 s8, s8
-; CHECK-NEXT:    vcvtb.f32.f16 s6, s6
 ; CHECK-NEXT:    vadd.f32 s0, s0, s8
-; CHECK-NEXT:    vcvtb.f32.f16 s4, s4
-; CHECK-NEXT:    vcvtb.f32.f16 s2, s2
 ; CHECK-NEXT:    vadd.f32 s0, s0, s6
 ; CHECK-NEXT:    vadd.f32 s0, s0, s4
 ; CHECK-NEXT:    vadd.f32 s0, s0, s2
@@ -1798,32 +1798,32 @@ define arm_aapcs_vfpcc float @half_short_mac(half* nocapture readonly %a, i16* n
 ; CHECK-NEXT:    @ =>This Inner Loop Header: Depth=1
 ; CHECK-NEXT:    ldrsh.w r4, [r3, #2]
 ; CHECK-NEXT:    vldr.16 s2, [r2, #2]
+; CHECK-NEXT:    ldrsh r5, [r3, #-2]
 ; CHECK-NEXT:    add.w r12, r12, #4
 ; CHECK-NEXT:    vmov s4, r4
-; CHECK-NEXT:    vcvt.f16.s32 s4, s4
 ; CHECK-NEXT:    ldrsh.w r4, [r3]
+; CHECK-NEXT:    vcvt.f16.s32 s4, s4
+; CHECK-NEXT:    vmov s8, r5
 ; CHECK-NEXT:    vmul.f16 s2, s2, s4
 ; CHECK-NEXT:    vldr.16 s4, [r2]
 ; CHECK-NEXT:    vmov s6, r4
-; CHECK-NEXT:    vcvt.f16.s32 s6, s6
-; CHECK-NEXT:    ldrsh r5, [r3, #-2]
 ; CHECK-NEXT:    ldrsh r4, [r3, #-4]
+; CHECK-NEXT:    vcvt.f16.s32 s6, s6
+; CHECK-NEXT:    vcvt.f16.s32 s8, s8
 ; CHECK-NEXT:    vmul.f16 s4, s4, s6
 ; CHECK-NEXT:    vldr.16 s6, [r2, #-2]
-; CHECK-NEXT:    adds r3, #8
-; CHECK-NEXT:    vmov s8, r5
-; CHECK-NEXT:    vcvt.f16.s32 s8, s8
 ; CHECK-NEXT:    vmov s10, r4
+; CHECK-NEXT:    vcvtb.f32.f16 s4, s4
 ; CHECK-NEXT:    vmul.f16 s6, s6, s8
 ; CHECK-NEXT:    vldr.16 s8, [r2, #-4]
 ; CHECK-NEXT:    vcvt.f16.s32 s10, s10
-; CHECK-NEXT:    adds r2, #8
+; CHECK-NEXT:    vcvtb.f32.f16 s6, s6
 ; CHECK-NEXT:    vmul.f16 s8, s8, s10
+; CHECK-NEXT:    vcvtb.f32.f16 s2, s2
 ; CHECK-NEXT:    vcvtb.f32.f16 s8, s8
-; CHECK-NEXT:    vcvtb.f32.f16 s6, s6
+; CHECK-NEXT:    adds r3, #8
 ; CHECK-NEXT:    vadd.f32 s0, s0, s8
-; CHECK-NEXT:    vcvtb.f32.f16 s4, s4
-; CHECK-NEXT:    vcvtb.f32.f16 s2, s2
+; CHECK-NEXT:    adds r2, #8
 ; CHECK-NEXT:    vadd.f32 s0, s0, s6
 ; CHECK-NEXT:    vadd.f32 s0, s0, s4
 ; CHECK-NEXT:    vadd.f32 s0, s0, s2

diff --git a/llvm/test/CodeGen/Thumb2/mve-masked-store.ll b/llvm/test/CodeGen/Thumb2/mve-masked-store.ll
index f2c51da..3a90529 100644 (file)
--- a/llvm/test/CodeGen/Thumb2/mve-masked-store.ll
+++ b/llvm/test/CodeGen/Thumb2/mve-masked-store.ll
@@ -1223,13 +1223,9 @@ define arm_aapcs_vfpcc void @masked_v4f16_align4(<4 x half> *%dest, <4 x float>
 ; CHECK-LE:       @ %bb.0: @ %entry
 ; CHECK-LE-NEXT:    .pad #4
 ; CHECK-LE-NEXT:    sub sp, #4
-; CHECK-LE-NEXT:    vcvtb.f16.f32 s4, s0
 ; CHECK-LE-NEXT:    vcmp.f32 s0, #0
-; CHECK-LE-NEXT:    vcvtt.f16.f32 s4, s1
+; CHECK-LE-NEXT:    movs r1, #0
 ; CHECK-LE-NEXT:    vmrs APSR_nzcv, fpscr
-; CHECK-LE-NEXT:    vcvtb.f16.f32 s5, s2
-; CHECK-LE-NEXT:    mov.w r1, #0
-; CHECK-LE-NEXT:    vcvtt.f16.f32 s5, s3
 ; CHECK-LE-NEXT:    it gt
 ; CHECK-LE-NEXT:    movgt r1, #1
 ; CHECK-LE-NEXT:    cmp r1, #0
@@ -1265,9 +1261,13 @@ define arm_aapcs_vfpcc void @masked_v4f16_align4(<4 x half> *%dest, <4 x float>
 ; CHECK-LE-NEXT:    cset r2, ne
 ; CHECK-LE-NEXT:    and r2, r2, #1
 ; CHECK-LE-NEXT:    rsbs r3, r3, #0
+; CHECK-LE-NEXT:    vcvtb.f16.f32 s4, s0
 ; CHECK-LE-NEXT:    bfi r1, r3, #2, #1
 ; CHECK-LE-NEXT:    rsbs r2, r2, #0
+; CHECK-LE-NEXT:    vcvtt.f16.f32 s4, s1
 ; CHECK-LE-NEXT:    bfi r1, r2, #3, #1
+; CHECK-LE-NEXT:    vcvtb.f16.f32 s5, s2
+; CHECK-LE-NEXT:    vcvtt.f16.f32 s5, s3
 ; CHECK-LE-NEXT:    lsls r2, r1, #31
 ; CHECK-LE-NEXT:    bne .LBB25_5
 ; CHECK-LE-NEXT:  @ %bb.1: @ %else
@@ -1307,17 +1307,13 @@ define arm_aapcs_vfpcc void @masked_v4f16_align4(<4 x half> *%dest, <4 x float>
 ; CHECK-BE-NEXT:    sub sp, #4
 ; CHECK-BE-NEXT:    vrev64.32 q1, q0
 ; CHECK-BE-NEXT:    movs r1, #0
-; CHECK-BE-NEXT:    vcvtb.f16.f32 s0, s4
 ; CHECK-BE-NEXT:    vcmp.f32 s4, #0
-; CHECK-BE-NEXT:    vcvtt.f16.f32 s0, s5
+; CHECK-BE-NEXT:    movs r2, #0
 ; CHECK-BE-NEXT:    vmrs APSR_nzcv, fpscr
-; CHECK-BE-NEXT:    vcvtb.f16.f32 s1, s6
-; CHECK-BE-NEXT:    vcmp.f32 s5, #0
-; CHECK-BE-NEXT:    vcvtt.f16.f32 s1, s7
 ; CHECK-BE-NEXT:    it gt
 ; CHECK-BE-NEXT:    movgt r1, #1
 ; CHECK-BE-NEXT:    cmp r1, #0
-; CHECK-BE-NEXT:    mov.w r2, #0
+; CHECK-BE-NEXT:    vcmp.f32 s5, #0
 ; CHECK-BE-NEXT:    cset r1, ne
 ; CHECK-BE-NEXT:    vmrs APSR_nzcv, fpscr
 ; CHECK-BE-NEXT:    and r1, r1, #1
@@ -1334,6 +1330,7 @@ define arm_aapcs_vfpcc void @masked_v4f16_align4(<4 x half> *%dest, <4 x float>
 ; CHECK-BE-NEXT:    and r3, r3, #1
 ; CHECK-BE-NEXT:    vcmp.f32 s7, #0
 ; CHECK-BE-NEXT:    rsb.w r3, r3, #0
+; CHECK-BE-NEXT:    vcvtb.f16.f32 s0, s4
 ; CHECK-BE-NEXT:    bfi r1, r3, #1, #1
 ; CHECK-BE-NEXT:    mov.w r3, #0
 ; CHECK-BE-NEXT:    it gt
@@ -1350,7 +1347,10 @@ define arm_aapcs_vfpcc void @masked_v4f16_align4(<4 x half> *%dest, <4 x float>
 ; CHECK-BE-NEXT:    rsbs r3, r3, #0
 ; CHECK-BE-NEXT:    bfi r1, r3, #2, #1
 ; CHECK-BE-NEXT:    rsbs r2, r2, #0
+; CHECK-BE-NEXT:    vcvtt.f16.f32 s0, s5
 ; CHECK-BE-NEXT:    bfi r1, r2, #3, #1
+; CHECK-BE-NEXT:    vcvtb.f16.f32 s1, s6
+; CHECK-BE-NEXT:    vcvtt.f16.f32 s1, s7
 ; CHECK-BE-NEXT:    lsls r2, r1, #31
 ; CHECK-BE-NEXT:    bne .LBB25_5
 ; CHECK-BE-NEXT:  @ %bb.1: @ %else
@@ -1395,13 +1395,9 @@ define arm_aapcs_vfpcc void @masked_v4f16_align2(<4 x half> *%dest, <4 x float>
 ; CHECK-LE:       @ %bb.0: @ %entry
 ; CHECK-LE-NEXT:    .pad #4
 ; CHECK-LE-NEXT:    sub sp, #4
-; CHECK-LE-NEXT:    vcvtb.f16.f32 s4, s0
 ; CHECK-LE-NEXT:    vcmp.f32 s0, #0
-; CHECK-LE-NEXT:    vcvtt.f16.f32 s4, s1
+; CHECK-LE-NEXT:    movs r1, #0
 ; CHECK-LE-NEXT:    vmrs APSR_nzcv, fpscr
-; CHECK-LE-NEXT:    vcvtb.f16.f32 s5, s2
-; CHECK-LE-NEXT:    mov.w r1, #0
-; CHECK-LE-NEXT:    vcvtt.f16.f32 s5, s3
 ; CHECK-LE-NEXT:    it gt
 ; CHECK-LE-NEXT:    movgt r1, #1
 ; CHECK-LE-NEXT:    cmp r1, #0
@@ -1437,9 +1433,13 @@ define arm_aapcs_vfpcc void @masked_v4f16_align2(<4 x half> *%dest, <4 x float>
 ; CHECK-LE-NEXT:    cset r2, ne
 ; CHECK-LE-NEXT:    and r2, r2, #1
 ; CHECK-LE-NEXT:    rsbs r3, r3, #0
+; CHECK-LE-NEXT:    vcvtb.f16.f32 s4, s0
 ; CHECK-LE-NEXT:    bfi r1, r3, #2, #1
 ; CHECK-LE-NEXT:    rsbs r2, r2, #0
+; CHECK-LE-NEXT:    vcvtt.f16.f32 s4, s1
 ; CHECK-LE-NEXT:    bfi r1, r2, #3, #1
+; CHECK-LE-NEXT:    vcvtb.f16.f32 s5, s2
+; CHECK-LE-NEXT:    vcvtt.f16.f32 s5, s3
 ; CHECK-LE-NEXT:    lsls r2, r1, #31
 ; CHECK-LE-NEXT:    bne .LBB26_5
 ; CHECK-LE-NEXT:  @ %bb.1: @ %else
@@ -1479,17 +1479,13 @@ define arm_aapcs_vfpcc void @masked_v4f16_align2(<4 x half> *%dest, <4 x float>
 ; CHECK-BE-NEXT:    sub sp, #4
 ; CHECK-BE-NEXT:    vrev64.32 q1, q0
 ; CHECK-BE-NEXT:    movs r1, #0
-; CHECK-BE-NEXT:    vcvtb.f16.f32 s0, s4
 ; CHECK-BE-NEXT:    vcmp.f32 s4, #0
-; CHECK-BE-NEXT:    vcvtt.f16.f32 s0, s5
+; CHECK-BE-NEXT:    movs r2, #0
 ; CHECK-BE-NEXT:    vmrs APSR_nzcv, fpscr
-; CHECK-BE-NEXT:    vcvtb.f16.f32 s1, s6
-; CHECK-BE-NEXT:    vcmp.f32 s5, #0
-; CHECK-BE-NEXT:    vcvtt.f16.f32 s1, s7
 ; CHECK-BE-NEXT:    it gt
 ; CHECK-BE-NEXT:    movgt r1, #1
 ; CHECK-BE-NEXT:    cmp r1, #0
-; CHECK-BE-NEXT:    mov.w r2, #0
+; CHECK-BE-NEXT:    vcmp.f32 s5, #0
 ; CHECK-BE-NEXT:    cset r1, ne
 ; CHECK-BE-NEXT:    vmrs APSR_nzcv, fpscr
 ; CHECK-BE-NEXT:    and r1, r1, #1
@@ -1506,6 +1502,7 @@ define arm_aapcs_vfpcc void @masked_v4f16_align2(<4 x half> *%dest, <4 x float>
 ; CHECK-BE-NEXT:    and r3, r3, #1
 ; CHECK-BE-NEXT:    vcmp.f32 s7, #0
 ; CHECK-BE-NEXT:    rsb.w r3, r3, #0
+; CHECK-BE-NEXT:    vcvtb.f16.f32 s0, s4
 ; CHECK-BE-NEXT:    bfi r1, r3, #1, #1
 ; CHECK-BE-NEXT:    mov.w r3, #0
 ; CHECK-BE-NEXT:    it gt
@@ -1522,7 +1519,10 @@ define arm_aapcs_vfpcc void @masked_v4f16_align2(<4 x half> *%dest, <4 x float>
 ; CHECK-BE-NEXT:    rsbs r3, r3, #0
 ; CHECK-BE-NEXT:    bfi r1, r3, #2, #1
 ; CHECK-BE-NEXT:    rsbs r2, r2, #0
+; CHECK-BE-NEXT:    vcvtt.f16.f32 s0, s5
 ; CHECK-BE-NEXT:    bfi r1, r2, #3, #1
+; CHECK-BE-NEXT:    vcvtb.f16.f32 s1, s6
+; CHECK-BE-NEXT:    vcvtt.f16.f32 s1, s7
 ; CHECK-BE-NEXT:    lsls r2, r1, #31
 ; CHECK-BE-NEXT:    bne .LBB26_5
 ; CHECK-BE-NEXT:  @ %bb.1: @ %else
@@ -1567,13 +1567,9 @@ define arm_aapcs_vfpcc void @masked_v4f16_align1(<4 x half> *%dest, <4 x float>
 ; CHECK-LE:       @ %bb.0: @ %entry
 ; CHECK-LE-NEXT:    .pad #20
 ; CHECK-LE-NEXT:    sub sp, #20
-; CHECK-LE-NEXT:    vcvtb.f16.f32 s4, s0
 ; CHECK-LE-NEXT:    vcmp.f32 s0, #0
-; CHECK-LE-NEXT:    vcvtt.f16.f32 s4, s1
+; CHECK-LE-NEXT:    movs r1, #0
 ; CHECK-LE-NEXT:    vmrs APSR_nzcv, fpscr
-; CHECK-LE-NEXT:    vcvtb.f16.f32 s5, s2
-; CHECK-LE-NEXT:    mov.w r1, #0
-; CHECK-LE-NEXT:    vcvtt.f16.f32 s5, s3
 ; CHECK-LE-NEXT:    it gt
 ; CHECK-LE-NEXT:    movgt r1, #1
 ; CHECK-LE-NEXT:    cmp r1, #0
@@ -1609,9 +1605,13 @@ define arm_aapcs_vfpcc void @masked_v4f16_align1(<4 x half> *%dest, <4 x float>
 ; CHECK-LE-NEXT:    cset r2, ne
 ; CHECK-LE-NEXT:    and r2, r2, #1
 ; CHECK-LE-NEXT:    rsbs r3, r3, #0
+; CHECK-LE-NEXT:    vcvtb.f16.f32 s4, s0
 ; CHECK-LE-NEXT:    bfi r1, r3, #2, #1
 ; CHECK-LE-NEXT:    rsbs r2, r2, #0
+; CHECK-LE-NEXT:    vcvtt.f16.f32 s4, s1
 ; CHECK-LE-NEXT:    bfi r1, r2, #3, #1
+; CHECK-LE-NEXT:    vcvtb.f16.f32 s5, s2
+; CHECK-LE-NEXT:    vcvtt.f16.f32 s5, s3
 ; CHECK-LE-NEXT:    lsls r2, r1, #31
 ; CHECK-LE-NEXT:    bne .LBB27_5
 ; CHECK-LE-NEXT:  @ %bb.1: @ %else
@@ -1659,17 +1659,13 @@ define arm_aapcs_vfpcc void @masked_v4f16_align1(<4 x half> *%dest, <4 x float>
 ; CHECK-BE-NEXT:    sub sp, #20
 ; CHECK-BE-NEXT:    vrev64.32 q1, q0
 ; CHECK-BE-NEXT:    movs r1, #0
-; CHECK-BE-NEXT:    vcvtb.f16.f32 s0, s4
 ; CHECK-BE-NEXT:    vcmp.f32 s4, #0
-; CHECK-BE-NEXT:    vcvtt.f16.f32 s0, s5
+; CHECK-BE-NEXT:    movs r2, #0
 ; CHECK-BE-NEXT:    vmrs APSR_nzcv, fpscr
-; CHECK-BE-NEXT:    vcvtb.f16.f32 s1, s6
-; CHECK-BE-NEXT:    vcmp.f32 s5, #0
-; CHECK-BE-NEXT:    vcvtt.f16.f32 s1, s7
 ; CHECK-BE-NEXT:    it gt
 ; CHECK-BE-NEXT:    movgt r1, #1
 ; CHECK-BE-NEXT:    cmp r1, #0
-; CHECK-BE-NEXT:    mov.w r2, #0
+; CHECK-BE-NEXT:    vcmp.f32 s5, #0
 ; CHECK-BE-NEXT:    cset r1, ne
 ; CHECK-BE-NEXT:    vmrs APSR_nzcv, fpscr
 ; CHECK-BE-NEXT:    and r1, r1, #1
@@ -1686,6 +1682,7 @@ define arm_aapcs_vfpcc void @masked_v4f16_align1(<4 x half> *%dest, <4 x float>
 ; CHECK-BE-NEXT:    and r3, r3, #1
 ; CHECK-BE-NEXT:    vcmp.f32 s7, #0
 ; CHECK-BE-NEXT:    rsb.w r3, r3, #0
+; CHECK-BE-NEXT:    vcvtb.f16.f32 s0, s4
 ; CHECK-BE-NEXT:    bfi r1, r3, #1, #1
 ; CHECK-BE-NEXT:    mov.w r3, #0
 ; CHECK-BE-NEXT:    it gt
@@ -1702,7 +1699,10 @@ define arm_aapcs_vfpcc void @masked_v4f16_align1(<4 x half> *%dest, <4 x float>
 ; CHECK-BE-NEXT:    rsbs r3, r3, #0
 ; CHECK-BE-NEXT:    bfi r1, r3, #2, #1
 ; CHECK-BE-NEXT:    rsbs r2, r2, #0
+; CHECK-BE-NEXT:    vcvtt.f16.f32 s0, s5
 ; CHECK-BE-NEXT:    bfi r1, r2, #3, #1
+; CHECK-BE-NEXT:    vcvtb.f16.f32 s1, s6
+; CHECK-BE-NEXT:    vcvtt.f16.f32 s1, s7
 ; CHECK-BE-NEXT:    lsls r2, r1, #31
 ; CHECK-BE-NEXT:    bne .LBB27_5
 ; CHECK-BE-NEXT:  @ %bb.1: @ %else

diff --git a/llvm/test/CodeGen/Thumb2/mve-vcvt.ll b/llvm/test/CodeGen/Thumb2/mve-vcvt.ll
index 0cfe5cc..ddc36f3 100644 (file)
--- a/llvm/test/CodeGen/Thumb2/mve-vcvt.ll
+++ b/llvm/test/CodeGen/Thumb2/mve-vcvt.ll
@@ -213,31 +213,31 @@ entry:
 define arm_aapcs_vfpcc <8 x i16> @foo_int16_half(<8 x half> %src) {
 ; CHECK-MVE-LABEL: foo_int16_half:
 ; CHECK-MVE:       @ %bb.0: @ %entry
+; CHECK-MVE-NEXT:    vmovx.f16 s14, s0
+; CHECK-MVE-NEXT:    vcvt.s32.f16 s0, s0
+; CHECK-MVE-NEXT:    vcvt.s32.f16 s14, s14
+; CHECK-MVE-NEXT:    vmov r0, s0
 ; CHECK-MVE-NEXT:    vmovx.f16 s4, s3
 ; CHECK-MVE-NEXT:    vmovx.f16 s6, s2
 ; CHECK-MVE-NEXT:    vmovx.f16 s10, s1
-; CHECK-MVE-NEXT:    vmovx.f16 s14, s0
-; CHECK-MVE-NEXT:    vcvt.s32.f16 s4, s4
-; CHECK-MVE-NEXT:    vcvt.s32.f16 s6, s6
 ; CHECK-MVE-NEXT:    vcvt.s32.f16 s8, s3
-; CHECK-MVE-NEXT:    vcvt.s32.f16 s10, s10
 ; CHECK-MVE-NEXT:    vcvt.s32.f16 s12, s2
-; CHECK-MVE-NEXT:    vcvt.s32.f16 s14, s14
 ; CHECK-MVE-NEXT:    vcvt.s32.f16 s5, s1
-; CHECK-MVE-NEXT:    vcvt.s32.f16 s0, s0
-; CHECK-MVE-NEXT:    vmov r0, s0
 ; CHECK-MVE-NEXT:    vmov.16 q0[0], r0
 ; CHECK-MVE-NEXT:    vmov r0, s14
 ; CHECK-MVE-NEXT:    vmov.16 q0[1], r0
 ; CHECK-MVE-NEXT:    vmov r0, s5
+; CHECK-MVE-NEXT:    vcvt.s32.f16 s10, s10
 ; CHECK-MVE-NEXT:    vmov.16 q0[2], r0
 ; CHECK-MVE-NEXT:    vmov r0, s10
+; CHECK-MVE-NEXT:    vcvt.s32.f16 s6, s6
 ; CHECK-MVE-NEXT:    vmov.16 q0[3], r0
 ; CHECK-MVE-NEXT:    vmov r0, s12
 ; CHECK-MVE-NEXT:    vmov.16 q0[4], r0
 ; CHECK-MVE-NEXT:    vmov r0, s6
 ; CHECK-MVE-NEXT:    vmov.16 q0[5], r0
 ; CHECK-MVE-NEXT:    vmov r0, s8
+; CHECK-MVE-NEXT:    vcvt.s32.f16 s4, s4
 ; CHECK-MVE-NEXT:    vmov.16 q0[6], r0
 ; CHECK-MVE-NEXT:    vmov r0, s4
 ; CHECK-MVE-NEXT:    vmov.16 q0[7], r0
@@ -255,31 +255,31 @@ entry:
 define arm_aapcs_vfpcc <8 x i16> @foo_uint16_half(<8 x half> %src) {
 ; CHECK-MVE-LABEL: foo_uint16_half:
 ; CHECK-MVE:       @ %bb.0: @ %entry
+; CHECK-MVE-NEXT:    vmovx.f16 s14, s0
+; CHECK-MVE-NEXT:    vcvt.s32.f16 s0, s0
+; CHECK-MVE-NEXT:    vcvt.s32.f16 s14, s14
+; CHECK-MVE-NEXT:    vmov r0, s0
 ; CHECK-MVE-NEXT:    vmovx.f16 s4, s3
 ; CHECK-MVE-NEXT:    vmovx.f16 s6, s2
 ; CHECK-MVE-NEXT:    vmovx.f16 s10, s1
-; CHECK-MVE-NEXT:    vmovx.f16 s14, s0
-; CHECK-MVE-NEXT:    vcvt.s32.f16 s4, s4
-; CHECK-MVE-NEXT:    vcvt.s32.f16 s6, s6
 ; CHECK-MVE-NEXT:    vcvt.s32.f16 s8, s3
-; CHECK-MVE-NEXT:    vcvt.s32.f16 s10, s10
 ; CHECK-MVE-NEXT:    vcvt.s32.f16 s12, s2
-; CHECK-MVE-NEXT:    vcvt.s32.f16 s14, s14
 ; CHECK-MVE-NEXT:    vcvt.s32.f16 s5, s1
-; CHECK-MVE-NEXT:    vcvt.s32.f16 s0, s0
-; CHECK-MVE-NEXT:    vmov r0, s0
 ; CHECK-MVE-NEXT:    vmov.16 q0[0], r0
 ; CHECK-MVE-NEXT:    vmov r0, s14
 ; CHECK-MVE-NEXT:    vmov.16 q0[1], r0
 ; CHECK-MVE-NEXT:    vmov r0, s5
+; CHECK-MVE-NEXT:    vcvt.s32.f16 s10, s10
 ; CHECK-MVE-NEXT:    vmov.16 q0[2], r0
 ; CHECK-MVE-NEXT:    vmov r0, s10
+; CHECK-MVE-NEXT:    vcvt.s32.f16 s6, s6
 ; CHECK-MVE-NEXT:    vmov.16 q0[3], r0
 ; CHECK-MVE-NEXT:    vmov r0, s12
 ; CHECK-MVE-NEXT:    vmov.16 q0[4], r0
 ; CHECK-MVE-NEXT:    vmov r0, s6
 ; CHECK-MVE-NEXT:    vmov.16 q0[5], r0
 ; CHECK-MVE-NEXT:    vmov r0, s8
+; CHECK-MVE-NEXT:    vcvt.s32.f16 s4, s4
 ; CHECK-MVE-NEXT:    vmov.16 q0[6], r0
 ; CHECK-MVE-NEXT:    vmov r0, s4
 ; CHECK-MVE-NEXT:    vmov.16 q0[7], r0

diff --git a/llvm/test/CodeGen/Thumb2/mve-vcvt16.ll b/llvm/test/CodeGen/Thumb2/mve-vcvt16.ll
index 86cc21d..c5421fe 100644 (file)
--- a/llvm/test/CodeGen/Thumb2/mve-vcvt16.ll
+++ b/llvm/test/CodeGen/Thumb2/mve-vcvt16.ll
@@ -18,15 +18,15 @@ entry:
 define arm_aapcs_vfpcc <8 x float> @fpext_8(<8 x half> %src1) {
 ; CHECK-LABEL: fpext_8:
 ; CHECK:       @ %bb.0: @ %entry
-; CHECK-NEXT:    vcvtt.f32.f16 s11, s1
-; CHECK-NEXT:    vcvtb.f32.f16 s10, s1
-; CHECK-NEXT:    vcvtt.f32.f16 s9, s0
-; CHECK-NEXT:    vcvtb.f32.f16 s8, s0
-; CHECK-NEXT:    vcvtt.f32.f16 s7, s3
-; CHECK-NEXT:    vcvtb.f32.f16 s6, s3
-; CHECK-NEXT:    vcvtt.f32.f16 s5, s2
-; CHECK-NEXT:    vcvtb.f32.f16 s4, s2
-; CHECK-NEXT:    vmov q0, q2
+; CHECK-NEXT:    vmov q2, q0
+; CHECK-NEXT:    vcvtt.f32.f16 s3, s9
+; CHECK-NEXT:    vcvtt.f32.f16 s7, s11
+; CHECK-NEXT:    vcvtb.f32.f16 s2, s9
+; CHECK-NEXT:    vcvtb.f32.f16 s6, s11
+; CHECK-NEXT:    vcvtt.f32.f16 s1, s8
+; CHECK-NEXT:    vcvtt.f32.f16 s5, s10
+; CHECK-NEXT:    vcvtb.f32.f16 s0, s8
+; CHECK-NEXT:    vcvtb.f32.f16 s4, s10
 ; CHECK-NEXT:    bx lr
 entry:
   %out = fpext <8 x half> %src1 to <8 x float>
@@ -247,12 +247,12 @@ define arm_aapcs_vfpcc <8 x float> @load_shuffleext_16(<16 x half>* %src) {
 ; CHECK-NEXT:    vld20.16 {q2, q3}, [r0]
 ; CHECK-NEXT:    vld21.16 {q2, q3}, [r0]
 ; CHECK-NEXT:    vcvtt.f32.f16 s3, s9
-; CHECK-NEXT:    vcvtb.f32.f16 s2, s9
-; CHECK-NEXT:    vcvtt.f32.f16 s1, s8
-; CHECK-NEXT:    vcvtb.f32.f16 s0, s8
 ; CHECK-NEXT:    vcvtt.f32.f16 s7, s11
+; CHECK-NEXT:    vcvtb.f32.f16 s2, s9
 ; CHECK-NEXT:    vcvtb.f32.f16 s6, s11
+; CHECK-NEXT:    vcvtt.f32.f16 s1, s8
 ; CHECK-NEXT:    vcvtt.f32.f16 s5, s10
+; CHECK-NEXT:    vcvtb.f32.f16 s0, s8
 ; CHECK-NEXT:    vcvtb.f32.f16 s4, s10
 ; CHECK-NEXT:    bx lr
 entry:

diff --git a/llvm/unittests/Target/ARM/MachineInstrTest.cpp b/llvm/unittests/Target/ARM/MachineInstrTest.cpp
index 51c9d73..3e5403f 100644 (file)
--- a/llvm/unittests/Target/ARM/MachineInstrTest.cpp
+++ b/llvm/unittests/Target/ARM/MachineInstrTest.cpp
@@ -973,13 +973,110 @@ TEST(MachineInstrValidTailPredication, IsCorrect) {
 TEST(MachineInstr, HasSideEffects) {
   using namespace ARM;
   std::set<unsigned> UnpredictableOpcodes = {
-      MVE_VCTP8,     MVE_VCTP16,    MVE_VCTP32,    MVE_VCTP64,    MVE_VPST,
-      MVE_VPTv16i8,  MVE_VPTv8i16,  MVE_VPTv4i32,  MVE_VPTv16i8r, MVE_VPTv8i16r,
-      MVE_VPTv4i32r, MVE_VPTv16s8,  MVE_VPTv8s16,  MVE_VPTv4s32,  MVE_VPTv16s8r,
-      MVE_VPTv8s16r, MVE_VPTv4s32r, MVE_VPTv16u8,  MVE_VPTv8u16,  MVE_VPTv4u32,
-      MVE_VPTv16u8r, MVE_VPTv8u16r, MVE_VPTv4u32r, MVE_VPTv8f16,  MVE_VPTv4f32,
-      MVE_VPTv8f16r, MVE_VPTv4f32r, MVE_VADC,      MVE_VADCI,     MVE_VSBC,
-      MVE_VSBCI,     MVE_VSHLC,
+      // MVE Instructions
+      MVE_VCTP8,
+      MVE_VCTP16,
+      MVE_VCTP32,
+      MVE_VCTP64,
+      MVE_VPST,
+      MVE_VPTv16i8,
+      MVE_VPTv8i16,
+      MVE_VPTv4i32,
+      MVE_VPTv16i8r,
+      MVE_VPTv8i16r,
+      MVE_VPTv4i32r,
+      MVE_VPTv16s8,
+      MVE_VPTv8s16,
+      MVE_VPTv4s32,
+      MVE_VPTv16s8r,
+      MVE_VPTv8s16r,
+      MVE_VPTv4s32r,
+      MVE_VPTv16u8,
+      MVE_VPTv8u16,
+      MVE_VPTv4u32,
+      MVE_VPTv16u8r,
+      MVE_VPTv8u16r,
+      MVE_VPTv4u32r,
+      MVE_VPTv8f16,
+      MVE_VPTv4f32,
+      MVE_VPTv8f16r,
+      MVE_VPTv4f32r,
+      MVE_VADC,
+      MVE_VADCI,
+      MVE_VSBC,
+      MVE_VSBCI,
+      MVE_VSHLC,
+      // FP Instructions
+      FLDMXIA,
+      FLDMXDB_UPD,
+      FLDMXIA_UPD,
+      FSTMXDB_UPD,
+      FSTMXIA,
+      FSTMXIA_UPD,
+      VLDR_FPCXTNS_off,
+      VLDR_FPCXTNS_off,
+      VLDR_FPCXTNS_post,
+      VLDR_FPCXTNS_pre,
+      VLDR_FPCXTS_off,
+      VLDR_FPCXTS_post,
+      VLDR_FPCXTS_pre,
+      VLDR_FPSCR_NZCVQC_off,
+      VLDR_FPSCR_NZCVQC_post,
+      VLDR_FPSCR_NZCVQC_pre,
+      VLDR_FPSCR_off,
+      VLDR_FPSCR_post,
+      VLDR_FPSCR_pre,
+      VLDR_P0_off,
+      VLDR_P0_post,
+      VLDR_P0_pre,
+      VLDR_VPR_off,
+      VLDR_VPR_post,
+      VLDR_VPR_pre,
+      VLLDM,
+      VLSTM,
+      VMRS,
+      VMRS_FPCXTNS,
+      VMRS_FPCXTS,
+      VMRS_FPEXC,
+      VMRS_FPINST,
+      VMRS_FPINST2,
+      VMRS_FPSCR_NZCVQC,
+      VMRS_FPSID,
+      VMRS_MVFR0,
+      VMRS_MVFR1,
+      VMRS_MVFR2,
+      VMRS_P0,
+      VMRS_VPR,
+      VMSR,
+      VMSR_FPCXTNS,
+      VMSR_FPCXTS,
+      VMSR_FPEXC,
+      VMSR_FPINST,
+      VMSR_FPINST2,
+      VMSR_FPSCR_NZCVQC,
+      VMSR_FPSID,
+      VMSR_P0,
+      VMSR_VPR,
+      VSCCLRMD,
+      VSCCLRMS,
+      VSTR_FPCXTNS_off,
+      VSTR_FPCXTNS_post,
+      VSTR_FPCXTNS_pre,
+      VSTR_FPCXTS_off,
+      VSTR_FPCXTS_post,
+      VSTR_FPCXTS_pre,
+      VSTR_FPSCR_NZCVQC_off,
+      VSTR_FPSCR_NZCVQC_post,
+      VSTR_FPSCR_NZCVQC_pre,
+      VSTR_FPSCR_off,
+      VSTR_FPSCR_post,
+      VSTR_FPSCR_pre,
+      VSTR_P0_off,
+      VSTR_P0_post,
+      VSTR_P0_pre,
+      VSTR_VPR_off,
+      VSTR_VPR_post,
+      VSTR_VPR_pre,
   };
 
   LLVMInitializeARMTargetInfo();
@@ -1006,7 +1103,8 @@ TEST(MachineInstr, HasSideEffects) {
 
   for (unsigned Op = 0; Op < ARM::INSTRUCTION_LIST_END; ++Op) {
     const MCInstrDesc &Desc = TII->get(Op);
-    if ((Desc.TSFlags & ARMII::DomainMask) != ARMII::DomainMVE)
+    if ((Desc.TSFlags & ARMII::DomainMask) != ARMII::DomainMVE &&
+        (Desc.TSFlags & ARMII::DomainMask) != ARMII::DomainVFP)
       continue;
     if (UnpredictableOpcodes.count(Op))
       continue;