[RISCV] Add VFCVT pseudos with no mask

When emitting a vfcvt with a rounding mode, we end up generating an unnecessary
vmset because the only rounding mode pseudos have a mask operand. This patch
adds a pseudo without a mask, and marks the masked variant with the
MaskedPseudo class so the doPeepholeMergeVMV optimisation knows to remove the
redundant vmset.

Reviewed By: reames

Differential Revision: https://reviews.llvm.org/D154266

diff --git a/llvm/lib/Target/RISCV/RISCVISelLowering.cpp b/llvm/lib/Target/RISCV/RISCVISelLowering.cpp
index 506b1c5..a66a3f0 100644 (file)
--- a/llvm/lib/Target/RISCV/RISCVISelLowering.cpp
+++ b/llvm/lib/Target/RISCV/RISCVISelLowering.cpp
@@ -13795,8 +13795,8 @@ static MachineBasicBlock *emitSelectPseudo(MachineInstr &MI,
   return TailMBB;
 }
 
-static MachineBasicBlock *
-emitVFCVT_RM_MASK(MachineInstr &MI, MachineBasicBlock *BB, unsigned Opcode) {
+static MachineBasicBlock *emitVFCVT_RM(MachineInstr &MI, MachineBasicBlock *BB,
+                                       unsigned Opcode) {
   DebugLoc DL = MI.getDebugLoc();
 
   const TargetInstrInfo &TII = *BB->getParent()->getSubtarget().getInstrInfo();
@@ -13804,20 +13804,20 @@ emitVFCVT_RM_MASK(MachineInstr &MI, MachineBasicBlock *BB, unsigned Opcode) {
   MachineRegisterInfo &MRI = BB->getParent()->getRegInfo();
   Register SavedFRM = MRI.createVirtualRegister(&RISCV::GPRRegClass);
 
+  assert(MI.getNumOperands() == 8 || MI.getNumOperands() == 7);
+  unsigned FRMIdx = MI.getNumOperands() == 8 ? 4 : 3;
+
   // Update FRM and save the old value.
   BuildMI(*BB, MI, DL, TII.get(RISCV::SwapFRMImm), SavedFRM)
-      .addImm(MI.getOperand(4).getImm());
+      .addImm(MI.getOperand(FRMIdx).getImm());
 
   // Emit an VFCVT without the FRM operand.
-  assert(MI.getNumOperands() == 8);
-  auto MIB = BuildMI(*BB, MI, DL, TII.get(Opcode))
-                 .add(MI.getOperand(0))
-                 .add(MI.getOperand(1))
-                 .add(MI.getOperand(2))
-                 .add(MI.getOperand(3))
-                 .add(MI.getOperand(5))
-                 .add(MI.getOperand(6))
-                 .add(MI.getOperand(7));
+  auto MIB = BuildMI(*BB, MI, DL, TII.get(Opcode));
+
+  for (unsigned I = 0; I < MI.getNumOperands(); I++)
+    if (I != FRMIdx)
+      MIB = MIB.add(MI.getOperand(I));
+
   if (MI.getFlag(MachineInstr::MIFlag::NoFPExcept))
     MIB->setFlag(MachineInstr::MIFlag::NoFPExcept);
 
@@ -14067,8 +14067,10 @@ RISCVTargetLowering::EmitInstrWithCustomInserter(MachineInstr &MI,
                          Subtarget);
 
 #define PseudoVFCVT_RM_LMUL_CASE(RMOpc, Opc, LMUL)                             \
+  case RISCV::RMOpc##_##LMUL:                                                  \
+    return emitVFCVT_RM(MI, BB, RISCV::Opc##_##LMUL);                          \
   case RISCV::RMOpc##_##LMUL##_MASK:                                           \
-    return emitVFCVT_RM_MASK(MI, BB, RISCV::Opc##_##LMUL##_MASK);
+    return emitVFCVT_RM(MI, BB, RISCV::Opc##_##LMUL##_MASK);
 
 #define PseudoVFCVT_RM_CASE(RMOpc, Opc)                                        \
   PseudoVFCVT_RM_LMUL_CASE(RMOpc, Opc, M1)                                     \

diff --git a/llvm/lib/Target/RISCV/RISCVInstrInfoVPseudos.td b/llvm/lib/Target/RISCV/RISCVInstrInfoVPseudos.td
index 795d3c9..b47b651 100644 (file)
--- a/llvm/lib/Target/RISCV/RISCVInstrInfoVPseudos.td
+++ b/llvm/lib/Target/RISCV/RISCVInstrInfoVPseudos.td
@@ -1054,6 +1054,20 @@ class VPseudoUnaryMask_NoExcept<VReg RetClass, VReg OpClass, string Constraint =
   let usesCustomInserter = 1;
 }
 
+class VPseudoUnaryNoMask_FRM<VReg RetClass, VReg OpClass, string Constraint = ""> :
+        Pseudo<(outs RetClass:$rd),
+               (ins RetClass:$merge, OpClass:$rs2, ixlenimm:$frm, AVL:$vl,
+                    ixlenimm:$sew, ixlenimm:$policy), []> {
+  let mayLoad = 0;
+  let mayStore = 0;
+  let hasSideEffects = 0;
+  let Constraints = !interleave([Constraint, "$rd = $merge"], ",");
+  let HasVLOp = 1;
+  let HasSEWOp = 1;
+  let HasVecPolicyOp = 1;
+  let usesCustomInserter = 1;
+}
+
 class VPseudoUnaryMask_FRM<VReg RetClass, VReg OpClass, string Constraint = ""> :
         Pseudo<(outs GetVRegNoV0<RetClass>.R:$rd),
                (ins GetVRegNoV0<RetClass>.R:$merge, OpClass:$rs2,
@@ -3515,8 +3529,13 @@ multiclass VPseudoConversionRM<VReg RetClass,
                                LMULInfo MInfo,
                                string Constraint = ""> {
   let VLMul = MInfo.value in {
-    def "_" # MInfo.MX # "_MASK" : VPseudoUnaryMask_FRM<RetClass, Op1Class,
+    def "_" # MInfo.MX : VPseudoUnaryNoMask_FRM<RetClass, Op1Class,
                                                         Constraint>;
+    def "_" # MInfo.MX # "_MASK" : VPseudoUnaryMask_FRM<RetClass, Op1Class,
+                                                        Constraint>,
+                                   RISCVMaskedPseudo</*MaskOpIdx*/ 2,
+                                                     /*HasTU*/ false,
+                                                     /*IsCombined*/true>;
   }
 }
 

diff --git a/llvm/test/CodeGen/RISCV/rvv/ctlz-sdnode.ll b/llvm/test/CodeGen/RISCV/rvv/ctlz-sdnode.ll
index 04fcebc..5e7aa20 100644 (file)
--- a/llvm/test/CodeGen/RISCV/rvv/ctlz-sdnode.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/ctlz-sdnode.ll
@@ -960,10 +960,9 @@ define <vscale x 1 x i32> @ctlz_nxv1i32(<vscale x 1 x i32> %va) {
 ;
 ; CHECK-F-LABEL: ctlz_nxv1i32:
 ; CHECK-F:       # %bb.0:
-; CHECK-F-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
-; CHECK-F-NEXT:    vmset.m v0
 ; CHECK-F-NEXT:    fsrmi a0, 1
-; CHECK-F-NEXT:    vfcvt.f.xu.v v8, v8, v0.t
+; CHECK-F-NEXT:    vsetvli a1, zero, e32, mf2, ta, ma
+; CHECK-F-NEXT:    vfcvt.f.xu.v v8, v8
 ; CHECK-F-NEXT:    vsrl.vi v8, v8, 23
 ; CHECK-F-NEXT:    li a1, 158
 ; CHECK-F-NEXT:    vrsub.vx v8, v8, a1
@@ -1066,10 +1065,9 @@ define <vscale x 2 x i32> @ctlz_nxv2i32(<vscale x 2 x i32> %va) {
 ;
 ; CHECK-F-LABEL: ctlz_nxv2i32:
 ; CHECK-F:       # %bb.0:
-; CHECK-F-NEXT:    vsetvli a0, zero, e32, m1, ta, ma
-; CHECK-F-NEXT:    vmset.m v0
 ; CHECK-F-NEXT:    fsrmi a0, 1
-; CHECK-F-NEXT:    vfcvt.f.xu.v v8, v8, v0.t
+; CHECK-F-NEXT:    vsetvli a1, zero, e32, m1, ta, ma
+; CHECK-F-NEXT:    vfcvt.f.xu.v v8, v8
 ; CHECK-F-NEXT:    vsrl.vi v8, v8, 23
 ; CHECK-F-NEXT:    li a1, 158
 ; CHECK-F-NEXT:    vrsub.vx v8, v8, a1
@@ -1172,10 +1170,9 @@ define <vscale x 4 x i32> @ctlz_nxv4i32(<vscale x 4 x i32> %va) {
 ;
 ; CHECK-F-LABEL: ctlz_nxv4i32:
 ; CHECK-F:       # %bb.0:
-; CHECK-F-NEXT:    vsetvli a0, zero, e32, m2, ta, ma
-; CHECK-F-NEXT:    vmset.m v0
 ; CHECK-F-NEXT:    fsrmi a0, 1
-; CHECK-F-NEXT:    vfcvt.f.xu.v v8, v8, v0.t
+; CHECK-F-NEXT:    vsetvli a1, zero, e32, m2, ta, ma
+; CHECK-F-NEXT:    vfcvt.f.xu.v v8, v8
 ; CHECK-F-NEXT:    vsrl.vi v8, v8, 23
 ; CHECK-F-NEXT:    li a1, 158
 ; CHECK-F-NEXT:    vrsub.vx v8, v8, a1
@@ -1278,10 +1275,9 @@ define <vscale x 8 x i32> @ctlz_nxv8i32(<vscale x 8 x i32> %va) {
 ;
 ; CHECK-F-LABEL: ctlz_nxv8i32:
 ; CHECK-F:       # %bb.0:
-; CHECK-F-NEXT:    vsetvli a0, zero, e32, m4, ta, ma
-; CHECK-F-NEXT:    vmset.m v0
 ; CHECK-F-NEXT:    fsrmi a0, 1
-; CHECK-F-NEXT:    vfcvt.f.xu.v v8, v8, v0.t
+; CHECK-F-NEXT:    vsetvli a1, zero, e32, m4, ta, ma
+; CHECK-F-NEXT:    vfcvt.f.xu.v v8, v8
 ; CHECK-F-NEXT:    vsrl.vi v8, v8, 23
 ; CHECK-F-NEXT:    li a1, 158
 ; CHECK-F-NEXT:    vrsub.vx v8, v8, a1
@@ -1384,10 +1380,9 @@ define <vscale x 16 x i32> @ctlz_nxv16i32(<vscale x 16 x i32> %va) {
 ;
 ; CHECK-F-LABEL: ctlz_nxv16i32:
 ; CHECK-F:       # %bb.0:
-; CHECK-F-NEXT:    vsetvli a0, zero, e32, m8, ta, ma
-; CHECK-F-NEXT:    vmset.m v0
 ; CHECK-F-NEXT:    fsrmi a0, 1
-; CHECK-F-NEXT:    vfcvt.f.xu.v v8, v8, v0.t
+; CHECK-F-NEXT:    vsetvli a1, zero, e32, m8, ta, ma
+; CHECK-F-NEXT:    vfcvt.f.xu.v v8, v8
 ; CHECK-F-NEXT:    vsrl.vi v8, v8, 23
 ; CHECK-F-NEXT:    li a1, 158
 ; CHECK-F-NEXT:    vrsub.vx v8, v8, a1
@@ -1398,10 +1393,9 @@ define <vscale x 16 x i32> @ctlz_nxv16i32(<vscale x 16 x i32> %va) {
 ;
 ; CHECK-D-LABEL: ctlz_nxv16i32:
 ; CHECK-D:       # %bb.0:
-; CHECK-D-NEXT:    vsetvli a0, zero, e32, m8, ta, ma
-; CHECK-D-NEXT:    vmset.m v0
 ; CHECK-D-NEXT:    fsrmi a0, 1
-; CHECK-D-NEXT:    vfcvt.f.xu.v v8, v8, v0.t
+; CHECK-D-NEXT:    vsetvli a1, zero, e32, m8, ta, ma
+; CHECK-D-NEXT:    vfcvt.f.xu.v v8, v8
 ; CHECK-D-NEXT:    vsrl.vi v8, v8, 23
 ; CHECK-D-NEXT:    li a1, 158
 ; CHECK-D-NEXT:    vrsub.vx v8, v8, a1
@@ -1524,10 +1518,9 @@ define <vscale x 1 x i64> @ctlz_nxv1i64(<vscale x 1 x i64> %va) {
 ;
 ; CHECK-F-LABEL: ctlz_nxv1i64:
 ; CHECK-F:       # %bb.0:
-; CHECK-F-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
-; CHECK-F-NEXT:    vmset.m v0
 ; CHECK-F-NEXT:    fsrmi a0, 1
-; CHECK-F-NEXT:    vfncvt.f.xu.w v9, v8, v0.t
+; CHECK-F-NEXT:    vsetvli a1, zero, e32, mf2, ta, ma
+; CHECK-F-NEXT:    vfncvt.f.xu.w v9, v8
 ; CHECK-F-NEXT:    vsrl.vi v8, v9, 23
 ; CHECK-F-NEXT:    vsetvli zero, zero, e64, m1, ta, ma
 ; CHECK-F-NEXT:    vzext.vf2 v9, v8
@@ -1540,10 +1533,9 @@ define <vscale x 1 x i64> @ctlz_nxv1i64(<vscale x 1 x i64> %va) {
 ;
 ; CHECK-D-LABEL: ctlz_nxv1i64:
 ; CHECK-D:       # %bb.0:
-; CHECK-D-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
-; CHECK-D-NEXT:    vmset.m v0
 ; CHECK-D-NEXT:    fsrmi a0, 1
-; CHECK-D-NEXT:    vfcvt.f.xu.v v8, v8, v0.t
+; CHECK-D-NEXT:    vsetvli a1, zero, e64, m1, ta, ma
+; CHECK-D-NEXT:    vfcvt.f.xu.v v8, v8
 ; CHECK-D-NEXT:    li a1, 52
 ; CHECK-D-NEXT:    vsrl.vx v8, v8, a1
 ; CHECK-D-NEXT:    li a1, 1086
@@ -1667,10 +1659,9 @@ define <vscale x 2 x i64> @ctlz_nxv2i64(<vscale x 2 x i64> %va) {
 ;
 ; CHECK-F-LABEL: ctlz_nxv2i64:
 ; CHECK-F:       # %bb.0:
-; CHECK-F-NEXT:    vsetvli a0, zero, e32, m1, ta, ma
-; CHECK-F-NEXT:    vmset.m v0
 ; CHECK-F-NEXT:    fsrmi a0, 1
-; CHECK-F-NEXT:    vfncvt.f.xu.w v10, v8, v0.t
+; CHECK-F-NEXT:    vsetvli a1, zero, e32, m1, ta, ma
+; CHECK-F-NEXT:    vfncvt.f.xu.w v10, v8
 ; CHECK-F-NEXT:    vsrl.vi v8, v10, 23
 ; CHECK-F-NEXT:    vsetvli zero, zero, e64, m2, ta, ma
 ; CHECK-F-NEXT:    vzext.vf2 v10, v8
@@ -1683,10 +1674,9 @@ define <vscale x 2 x i64> @ctlz_nxv2i64(<vscale x 2 x i64> %va) {
 ;
 ; CHECK-D-LABEL: ctlz_nxv2i64:
 ; CHECK-D:       # %bb.0:
-; CHECK-D-NEXT:    vsetvli a0, zero, e64, m2, ta, ma
-; CHECK-D-NEXT:    vmset.m v0
 ; CHECK-D-NEXT:    fsrmi a0, 1
-; CHECK-D-NEXT:    vfcvt.f.xu.v v8, v8, v0.t
+; CHECK-D-NEXT:    vsetvli a1, zero, e64, m2, ta, ma
+; CHECK-D-NEXT:    vfcvt.f.xu.v v8, v8
 ; CHECK-D-NEXT:    li a1, 52
 ; CHECK-D-NEXT:    vsrl.vx v8, v8, a1
 ; CHECK-D-NEXT:    li a1, 1086
@@ -1810,10 +1800,9 @@ define <vscale x 4 x i64> @ctlz_nxv4i64(<vscale x 4 x i64> %va) {
 ;
 ; CHECK-F-LABEL: ctlz_nxv4i64:
 ; CHECK-F:       # %bb.0:
-; CHECK-F-NEXT:    vsetvli a0, zero, e32, m2, ta, ma
-; CHECK-F-NEXT:    vmset.m v0
 ; CHECK-F-NEXT:    fsrmi a0, 1
-; CHECK-F-NEXT:    vfncvt.f.xu.w v12, v8, v0.t
+; CHECK-F-NEXT:    vsetvli a1, zero, e32, m2, ta, ma
+; CHECK-F-NEXT:    vfncvt.f.xu.w v12, v8
 ; CHECK-F-NEXT:    vsrl.vi v8, v12, 23
 ; CHECK-F-NEXT:    vsetvli zero, zero, e64, m4, ta, ma
 ; CHECK-F-NEXT:    vzext.vf2 v12, v8
@@ -1826,10 +1815,9 @@ define <vscale x 4 x i64> @ctlz_nxv4i64(<vscale x 4 x i64> %va) {
 ;
 ; CHECK-D-LABEL: ctlz_nxv4i64:
 ; CHECK-D:       # %bb.0:
-; CHECK-D-NEXT:    vsetvli a0, zero, e64, m4, ta, ma
-; CHECK-D-NEXT:    vmset.m v0
 ; CHECK-D-NEXT:    fsrmi a0, 1
-; CHECK-D-NEXT:    vfcvt.f.xu.v v8, v8, v0.t
+; CHECK-D-NEXT:    vsetvli a1, zero, e64, m4, ta, ma
+; CHECK-D-NEXT:    vfcvt.f.xu.v v8, v8
 ; CHECK-D-NEXT:    li a1, 52
 ; CHECK-D-NEXT:    vsrl.vx v8, v8, a1
 ; CHECK-D-NEXT:    li a1, 1086
@@ -1953,10 +1941,9 @@ define <vscale x 8 x i64> @ctlz_nxv8i64(<vscale x 8 x i64> %va) {
 ;
 ; CHECK-F-LABEL: ctlz_nxv8i64:
 ; CHECK-F:       # %bb.0:
-; CHECK-F-NEXT:    vsetvli a0, zero, e32, m4, ta, ma
-; CHECK-F-NEXT:    vmset.m v0
 ; CHECK-F-NEXT:    fsrmi a0, 1
-; CHECK-F-NEXT:    vfncvt.f.xu.w v16, v8, v0.t
+; CHECK-F-NEXT:    vsetvli a1, zero, e32, m4, ta, ma
+; CHECK-F-NEXT:    vfncvt.f.xu.w v16, v8
 ; CHECK-F-NEXT:    vsrl.vi v8, v16, 23
 ; CHECK-F-NEXT:    vsetvli zero, zero, e64, m8, ta, ma
 ; CHECK-F-NEXT:    vzext.vf2 v16, v8
@@ -1969,10 +1956,9 @@ define <vscale x 8 x i64> @ctlz_nxv8i64(<vscale x 8 x i64> %va) {
 ;
 ; CHECK-D-LABEL: ctlz_nxv8i64:
 ; CHECK-D:       # %bb.0:
-; CHECK-D-NEXT:    vsetvli a0, zero, e64, m8, ta, ma
-; CHECK-D-NEXT:    vmset.m v0
 ; CHECK-D-NEXT:    fsrmi a0, 1
-; CHECK-D-NEXT:    vfcvt.f.xu.v v8, v8, v0.t
+; CHECK-D-NEXT:    vsetvli a1, zero, e64, m8, ta, ma
+; CHECK-D-NEXT:    vfcvt.f.xu.v v8, v8
 ; CHECK-D-NEXT:    li a1, 52
 ; CHECK-D-NEXT:    vsrl.vx v8, v8, a1
 ; CHECK-D-NEXT:    li a1, 1086
@@ -2887,10 +2873,9 @@ define <vscale x 1 x i32> @ctlz_zero_undef_nxv1i32(<vscale x 1 x i32> %va) {
 ;
 ; CHECK-F-LABEL: ctlz_zero_undef_nxv1i32:
 ; CHECK-F:       # %bb.0:
-; CHECK-F-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
-; CHECK-F-NEXT:    vmset.m v0
 ; CHECK-F-NEXT:    fsrmi a0, 1
-; CHECK-F-NEXT:    vfcvt.f.xu.v v8, v8, v0.t
+; CHECK-F-NEXT:    vsetvli a1, zero, e32, mf2, ta, ma
+; CHECK-F-NEXT:    vfcvt.f.xu.v v8, v8
 ; CHECK-F-NEXT:    vsrl.vi v8, v8, 23
 ; CHECK-F-NEXT:    li a1, 158
 ; CHECK-F-NEXT:    vrsub.vx v8, v8, a1
@@ -2988,10 +2973,9 @@ define <vscale x 2 x i32> @ctlz_zero_undef_nxv2i32(<vscale x 2 x i32> %va) {
 ;
 ; CHECK-F-LABEL: ctlz_zero_undef_nxv2i32:
 ; CHECK-F:       # %bb.0:
-; CHECK-F-NEXT:    vsetvli a0, zero, e32, m1, ta, ma
-; CHECK-F-NEXT:    vmset.m v0
 ; CHECK-F-NEXT:    fsrmi a0, 1
-; CHECK-F-NEXT:    vfcvt.f.xu.v v8, v8, v0.t
+; CHECK-F-NEXT:    vsetvli a1, zero, e32, m1, ta, ma
+; CHECK-F-NEXT:    vfcvt.f.xu.v v8, v8
 ; CHECK-F-NEXT:    vsrl.vi v8, v8, 23
 ; CHECK-F-NEXT:    li a1, 158
 ; CHECK-F-NEXT:    vrsub.vx v8, v8, a1
@@ -3089,10 +3073,9 @@ define <vscale x 4 x i32> @ctlz_zero_undef_nxv4i32(<vscale x 4 x i32> %va) {
 ;
 ; CHECK-F-LABEL: ctlz_zero_undef_nxv4i32:
 ; CHECK-F:       # %bb.0:
-; CHECK-F-NEXT:    vsetvli a0, zero, e32, m2, ta, ma
-; CHECK-F-NEXT:    vmset.m v0
 ; CHECK-F-NEXT:    fsrmi a0, 1
-; CHECK-F-NEXT:    vfcvt.f.xu.v v8, v8, v0.t
+; CHECK-F-NEXT:    vsetvli a1, zero, e32, m2, ta, ma
+; CHECK-F-NEXT:    vfcvt.f.xu.v v8, v8
 ; CHECK-F-NEXT:    vsrl.vi v8, v8, 23
 ; CHECK-F-NEXT:    li a1, 158
 ; CHECK-F-NEXT:    vrsub.vx v8, v8, a1
@@ -3190,10 +3173,9 @@ define <vscale x 8 x i32> @ctlz_zero_undef_nxv8i32(<vscale x 8 x i32> %va) {
 ;
 ; CHECK-F-LABEL: ctlz_zero_undef_nxv8i32:
 ; CHECK-F:       # %bb.0:
-; CHECK-F-NEXT:    vsetvli a0, zero, e32, m4, ta, ma
-; CHECK-F-NEXT:    vmset.m v0
 ; CHECK-F-NEXT:    fsrmi a0, 1
-; CHECK-F-NEXT:    vfcvt.f.xu.v v8, v8, v0.t
+; CHECK-F-NEXT:    vsetvli a1, zero, e32, m4, ta, ma
+; CHECK-F-NEXT:    vfcvt.f.xu.v v8, v8
 ; CHECK-F-NEXT:    vsrl.vi v8, v8, 23
 ; CHECK-F-NEXT:    li a1, 158
 ; CHECK-F-NEXT:    vrsub.vx v8, v8, a1
@@ -3291,10 +3273,9 @@ define <vscale x 16 x i32> @ctlz_zero_undef_nxv16i32(<vscale x 16 x i32> %va) {
 ;
 ; CHECK-F-LABEL: ctlz_zero_undef_nxv16i32:
 ; CHECK-F:       # %bb.0:
-; CHECK-F-NEXT:    vsetvli a0, zero, e32, m8, ta, ma
-; CHECK-F-NEXT:    vmset.m v0
 ; CHECK-F-NEXT:    fsrmi a0, 1
-; CHECK-F-NEXT:    vfcvt.f.xu.v v8, v8, v0.t
+; CHECK-F-NEXT:    vsetvli a1, zero, e32, m8, ta, ma
+; CHECK-F-NEXT:    vfcvt.f.xu.v v8, v8
 ; CHECK-F-NEXT:    vsrl.vi v8, v8, 23
 ; CHECK-F-NEXT:    li a1, 158
 ; CHECK-F-NEXT:    vrsub.vx v8, v8, a1
@@ -3303,10 +3284,9 @@ define <vscale x 16 x i32> @ctlz_zero_undef_nxv16i32(<vscale x 16 x i32> %va) {
 ;
 ; CHECK-D-LABEL: ctlz_zero_undef_nxv16i32:
 ; CHECK-D:       # %bb.0:
-; CHECK-D-NEXT:    vsetvli a0, zero, e32, m8, ta, ma
-; CHECK-D-NEXT:    vmset.m v0
 ; CHECK-D-NEXT:    fsrmi a0, 1
-; CHECK-D-NEXT:    vfcvt.f.xu.v v8, v8, v0.t
+; CHECK-D-NEXT:    vsetvli a1, zero, e32, m8, ta, ma
+; CHECK-D-NEXT:    vfcvt.f.xu.v v8, v8
 ; CHECK-D-NEXT:    vsrl.vi v8, v8, 23
 ; CHECK-D-NEXT:    li a1, 158
 ; CHECK-D-NEXT:    vrsub.vx v8, v8, a1
@@ -3426,10 +3406,9 @@ define <vscale x 1 x i64> @ctlz_zero_undef_nxv1i64(<vscale x 1 x i64> %va) {
 ;
 ; CHECK-F-LABEL: ctlz_zero_undef_nxv1i64:
 ; CHECK-F:       # %bb.0:
-; CHECK-F-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
-; CHECK-F-NEXT:    vmset.m v0
 ; CHECK-F-NEXT:    fsrmi a0, 1
-; CHECK-F-NEXT:    vfncvt.f.xu.w v9, v8, v0.t
+; CHECK-F-NEXT:    vsetvli a1, zero, e32, mf2, ta, ma
+; CHECK-F-NEXT:    vfncvt.f.xu.w v9, v8
 ; CHECK-F-NEXT:    vsrl.vi v8, v9, 23
 ; CHECK-F-NEXT:    vsetvli zero, zero, e64, m1, ta, ma
 ; CHECK-F-NEXT:    vzext.vf2 v9, v8
@@ -3440,10 +3419,9 @@ define <vscale x 1 x i64> @ctlz_zero_undef_nxv1i64(<vscale x 1 x i64> %va) {
 ;
 ; CHECK-D-LABEL: ctlz_zero_undef_nxv1i64:
 ; CHECK-D:       # %bb.0:
-; CHECK-D-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
-; CHECK-D-NEXT:    vmset.m v0
 ; CHECK-D-NEXT:    fsrmi a0, 1
-; CHECK-D-NEXT:    vfcvt.f.xu.v v8, v8, v0.t
+; CHECK-D-NEXT:    vsetvli a1, zero, e64, m1, ta, ma
+; CHECK-D-NEXT:    vfcvt.f.xu.v v8, v8
 ; CHECK-D-NEXT:    li a1, 52
 ; CHECK-D-NEXT:    vsrl.vx v8, v8, a1
 ; CHECK-D-NEXT:    li a1, 1086
@@ -3564,10 +3542,9 @@ define <vscale x 2 x i64> @ctlz_zero_undef_nxv2i64(<vscale x 2 x i64> %va) {
 ;
 ; CHECK-F-LABEL: ctlz_zero_undef_nxv2i64:
 ; CHECK-F:       # %bb.0:
-; CHECK-F-NEXT:    vsetvli a0, zero, e32, m1, ta, ma
-; CHECK-F-NEXT:    vmset.m v0
 ; CHECK-F-NEXT:    fsrmi a0, 1
-; CHECK-F-NEXT:    vfncvt.f.xu.w v10, v8, v0.t
+; CHECK-F-NEXT:    vsetvli a1, zero, e32, m1, ta, ma
+; CHECK-F-NEXT:    vfncvt.f.xu.w v10, v8
 ; CHECK-F-NEXT:    vsrl.vi v8, v10, 23
 ; CHECK-F-NEXT:    vsetvli zero, zero, e64, m2, ta, ma
 ; CHECK-F-NEXT:    vzext.vf2 v10, v8
@@ -3578,10 +3555,9 @@ define <vscale x 2 x i64> @ctlz_zero_undef_nxv2i64(<vscale x 2 x i64> %va) {
 ;
 ; CHECK-D-LABEL: ctlz_zero_undef_nxv2i64:
 ; CHECK-D:       # %bb.0:
-; CHECK-D-NEXT:    vsetvli a0, zero, e64, m2, ta, ma
-; CHECK-D-NEXT:    vmset.m v0
 ; CHECK-D-NEXT:    fsrmi a0, 1
-; CHECK-D-NEXT:    vfcvt.f.xu.v v8, v8, v0.t
+; CHECK-D-NEXT:    vsetvli a1, zero, e64, m2, ta, ma
+; CHECK-D-NEXT:    vfcvt.f.xu.v v8, v8
 ; CHECK-D-NEXT:    li a1, 52
 ; CHECK-D-NEXT:    vsrl.vx v8, v8, a1
 ; CHECK-D-NEXT:    li a1, 1086
@@ -3702,10 +3678,9 @@ define <vscale x 4 x i64> @ctlz_zero_undef_nxv4i64(<vscale x 4 x i64> %va) {
 ;
 ; CHECK-F-LABEL: ctlz_zero_undef_nxv4i64:
 ; CHECK-F:       # %bb.0:
-; CHECK-F-NEXT:    vsetvli a0, zero, e32, m2, ta, ma
-; CHECK-F-NEXT:    vmset.m v0
 ; CHECK-F-NEXT:    fsrmi a0, 1
-; CHECK-F-NEXT:    vfncvt.f.xu.w v12, v8, v0.t
+; CHECK-F-NEXT:    vsetvli a1, zero, e32, m2, ta, ma
+; CHECK-F-NEXT:    vfncvt.f.xu.w v12, v8
 ; CHECK-F-NEXT:    vsrl.vi v8, v12, 23
 ; CHECK-F-NEXT:    vsetvli zero, zero, e64, m4, ta, ma
 ; CHECK-F-NEXT:    vzext.vf2 v12, v8
@@ -3716,10 +3691,9 @@ define <vscale x 4 x i64> @ctlz_zero_undef_nxv4i64(<vscale x 4 x i64> %va) {
 ;
 ; CHECK-D-LABEL: ctlz_zero_undef_nxv4i64:
 ; CHECK-D:       # %bb.0:
-; CHECK-D-NEXT:    vsetvli a0, zero, e64, m4, ta, ma
-; CHECK-D-NEXT:    vmset.m v0
 ; CHECK-D-NEXT:    fsrmi a0, 1
-; CHECK-D-NEXT:    vfcvt.f.xu.v v8, v8, v0.t
+; CHECK-D-NEXT:    vsetvli a1, zero, e64, m4, ta, ma
+; CHECK-D-NEXT:    vfcvt.f.xu.v v8, v8
 ; CHECK-D-NEXT:    li a1, 52
 ; CHECK-D-NEXT:    vsrl.vx v8, v8, a1
 ; CHECK-D-NEXT:    li a1, 1086
@@ -3840,10 +3814,9 @@ define <vscale x 8 x i64> @ctlz_zero_undef_nxv8i64(<vscale x 8 x i64> %va) {
 ;
 ; CHECK-F-LABEL: ctlz_zero_undef_nxv8i64:
 ; CHECK-F:       # %bb.0:
-; CHECK-F-NEXT:    vsetvli a0, zero, e32, m4, ta, ma
-; CHECK-F-NEXT:    vmset.m v0
 ; CHECK-F-NEXT:    fsrmi a0, 1
-; CHECK-F-NEXT:    vfncvt.f.xu.w v16, v8, v0.t
+; CHECK-F-NEXT:    vsetvli a1, zero, e32, m4, ta, ma
+; CHECK-F-NEXT:    vfncvt.f.xu.w v16, v8
 ; CHECK-F-NEXT:    vsrl.vi v8, v16, 23
 ; CHECK-F-NEXT:    vsetvli zero, zero, e64, m8, ta, ma
 ; CHECK-F-NEXT:    vzext.vf2 v16, v8
@@ -3854,10 +3827,9 @@ define <vscale x 8 x i64> @ctlz_zero_undef_nxv8i64(<vscale x 8 x i64> %va) {
 ;
 ; CHECK-D-LABEL: ctlz_zero_undef_nxv8i64:
 ; CHECK-D:       # %bb.0:
-; CHECK-D-NEXT:    vsetvli a0, zero, e64, m8, ta, ma
-; CHECK-D-NEXT:    vmset.m v0
 ; CHECK-D-NEXT:    fsrmi a0, 1
-; CHECK-D-NEXT:    vfcvt.f.xu.v v8, v8, v0.t
+; CHECK-D-NEXT:    vsetvli a1, zero, e64, m8, ta, ma
+; CHECK-D-NEXT:    vfcvt.f.xu.v v8, v8
 ; CHECK-D-NEXT:    li a1, 52
 ; CHECK-D-NEXT:    vsrl.vx v8, v8, a1
 ; CHECK-D-NEXT:    li a1, 1086

diff --git a/llvm/test/CodeGen/RISCV/rvv/cttz-sdnode.ll b/llvm/test/CodeGen/RISCV/rvv/cttz-sdnode.ll
index 77d0183..c80a0e0 100644 (file)
--- a/llvm/test/CodeGen/RISCV/rvv/cttz-sdnode.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/cttz-sdnode.ll
@@ -928,9 +928,8 @@ define <vscale x 1 x i32> @cttz_nxv1i32(<vscale x 1 x i32> %va) {
 ; CHECK-F-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
 ; CHECK-F-NEXT:    vrsub.vi v9, v8, 0
 ; CHECK-F-NEXT:    vand.vv v9, v8, v9
-; CHECK-F-NEXT:    vmset.m v0
 ; CHECK-F-NEXT:    fsrmi a0, 1
-; CHECK-F-NEXT:    vfcvt.f.xu.v v9, v9, v0.t
+; CHECK-F-NEXT:    vfcvt.f.xu.v v9, v9
 ; CHECK-F-NEXT:    vsrl.vi v9, v9, 23
 ; CHECK-F-NEXT:    li a1, 127
 ; CHECK-F-NEXT:    vsub.vx v9, v9, a1
@@ -1026,9 +1025,8 @@ define <vscale x 2 x i32> @cttz_nxv2i32(<vscale x 2 x i32> %va) {
 ; CHECK-F-NEXT:    vsetvli a0, zero, e32, m1, ta, ma
 ; CHECK-F-NEXT:    vrsub.vi v9, v8, 0
 ; CHECK-F-NEXT:    vand.vv v9, v8, v9
-; CHECK-F-NEXT:    vmset.m v0
 ; CHECK-F-NEXT:    fsrmi a0, 1
-; CHECK-F-NEXT:    vfcvt.f.xu.v v9, v9, v0.t
+; CHECK-F-NEXT:    vfcvt.f.xu.v v9, v9
 ; CHECK-F-NEXT:    vsrl.vi v9, v9, 23
 ; CHECK-F-NEXT:    li a1, 127
 ; CHECK-F-NEXT:    vsub.vx v9, v9, a1
@@ -1124,9 +1122,8 @@ define <vscale x 4 x i32> @cttz_nxv4i32(<vscale x 4 x i32> %va) {
 ; CHECK-F-NEXT:    vsetvli a0, zero, e32, m2, ta, ma
 ; CHECK-F-NEXT:    vrsub.vi v10, v8, 0
 ; CHECK-F-NEXT:    vand.vv v10, v8, v10
-; CHECK-F-NEXT:    vmset.m v0
 ; CHECK-F-NEXT:    fsrmi a0, 1
-; CHECK-F-NEXT:    vfcvt.f.xu.v v10, v10, v0.t
+; CHECK-F-NEXT:    vfcvt.f.xu.v v10, v10
 ; CHECK-F-NEXT:    vsrl.vi v10, v10, 23
 ; CHECK-F-NEXT:    li a1, 127
 ; CHECK-F-NEXT:    vsub.vx v10, v10, a1
@@ -1222,9 +1219,8 @@ define <vscale x 8 x i32> @cttz_nxv8i32(<vscale x 8 x i32> %va) {
 ; CHECK-F-NEXT:    vsetvli a0, zero, e32, m4, ta, ma
 ; CHECK-F-NEXT:    vrsub.vi v12, v8, 0
 ; CHECK-F-NEXT:    vand.vv v12, v8, v12
-; CHECK-F-NEXT:    vmset.m v0
 ; CHECK-F-NEXT:    fsrmi a0, 1
-; CHECK-F-NEXT:    vfcvt.f.xu.v v12, v12, v0.t
+; CHECK-F-NEXT:    vfcvt.f.xu.v v12, v12
 ; CHECK-F-NEXT:    vsrl.vi v12, v12, 23
 ; CHECK-F-NEXT:    li a1, 127
 ; CHECK-F-NEXT:    vsub.vx v12, v12, a1
@@ -1320,9 +1316,8 @@ define <vscale x 16 x i32> @cttz_nxv16i32(<vscale x 16 x i32> %va) {
 ; CHECK-F-NEXT:    vsetvli a0, zero, e32, m8, ta, ma
 ; CHECK-F-NEXT:    vrsub.vi v16, v8, 0
 ; CHECK-F-NEXT:    vand.vv v16, v8, v16
-; CHECK-F-NEXT:    vmset.m v0
 ; CHECK-F-NEXT:    fsrmi a0, 1
-; CHECK-F-NEXT:    vfcvt.f.xu.v v16, v16, v0.t
+; CHECK-F-NEXT:    vfcvt.f.xu.v v16, v16
 ; CHECK-F-NEXT:    vsrl.vi v16, v16, 23
 ; CHECK-F-NEXT:    li a1, 127
 ; CHECK-F-NEXT:    vsub.vx v16, v16, a1
@@ -1337,9 +1332,8 @@ define <vscale x 16 x i32> @cttz_nxv16i32(<vscale x 16 x i32> %va) {
 ; CHECK-D-NEXT:    vsetvli a0, zero, e32, m8, ta, ma
 ; CHECK-D-NEXT:    vrsub.vi v16, v8, 0
 ; CHECK-D-NEXT:    vand.vv v16, v8, v16
-; CHECK-D-NEXT:    vmset.m v0
 ; CHECK-D-NEXT:    fsrmi a0, 1
-; CHECK-D-NEXT:    vfcvt.f.xu.v v16, v16, v0.t
+; CHECK-D-NEXT:    vfcvt.f.xu.v v16, v16
 ; CHECK-D-NEXT:    vsrl.vi v16, v16, 23
 ; CHECK-D-NEXT:    li a1, 127
 ; CHECK-D-NEXT:    vsub.vx v16, v16, a1
@@ -1446,10 +1440,9 @@ define <vscale x 1 x i64> @cttz_nxv1i64(<vscale x 1 x i64> %va) {
 ; RV32F-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
 ; RV32F-NEXT:    vrsub.vi v9, v8, 0
 ; RV32F-NEXT:    vand.vv v9, v8, v9
-; RV32F-NEXT:    vmset.m v0
 ; RV32F-NEXT:    fsrmi a0, 1
 ; RV32F-NEXT:    vsetvli zero, zero, e32, mf2, ta, ma
-; RV32F-NEXT:    vfncvt.f.xu.w v10, v9, v0.t
+; RV32F-NEXT:    vfncvt.f.xu.w v10, v9
 ; RV32F-NEXT:    vsrl.vi v9, v10, 23
 ; RV32F-NEXT:    vsetvli zero, zero, e64, m1, ta, ma
 ; RV32F-NEXT:    vzext.vf2 v10, v9
@@ -1466,10 +1459,9 @@ define <vscale x 1 x i64> @cttz_nxv1i64(<vscale x 1 x i64> %va) {
 ; RV64F-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
 ; RV64F-NEXT:    vrsub.vi v9, v8, 0
 ; RV64F-NEXT:    vand.vv v9, v8, v9
-; RV64F-NEXT:    vmset.m v0
 ; RV64F-NEXT:    fsrmi a0, 1
 ; RV64F-NEXT:    vsetvli zero, zero, e32, mf2, ta, ma
-; RV64F-NEXT:    vfncvt.f.xu.w v10, v9, v0.t
+; RV64F-NEXT:    vfncvt.f.xu.w v10, v9
 ; RV64F-NEXT:    vsrl.vi v9, v10, 23
 ; RV64F-NEXT:    vsetvli zero, zero, e64, m1, ta, ma
 ; RV64F-NEXT:    vzext.vf2 v10, v9
@@ -1486,9 +1478,8 @@ define <vscale x 1 x i64> @cttz_nxv1i64(<vscale x 1 x i64> %va) {
 ; RV32D-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
 ; RV32D-NEXT:    vrsub.vi v9, v8, 0
 ; RV32D-NEXT:    vand.vv v9, v8, v9
-; RV32D-NEXT:    vmset.m v0
 ; RV32D-NEXT:    fsrmi a0, 1
-; RV32D-NEXT:    vfcvt.f.xu.v v9, v9, v0.t
+; RV32D-NEXT:    vfcvt.f.xu.v v9, v9
 ; RV32D-NEXT:    li a1, 52
 ; RV32D-NEXT:    vsrl.vx v9, v9, a1
 ; RV32D-NEXT:    li a1, 1023
@@ -1504,9 +1495,8 @@ define <vscale x 1 x i64> @cttz_nxv1i64(<vscale x 1 x i64> %va) {
 ; RV64D-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
 ; RV64D-NEXT:    vrsub.vi v9, v8, 0
 ; RV64D-NEXT:    vand.vv v9, v8, v9
-; RV64D-NEXT:    vmset.m v0
 ; RV64D-NEXT:    fsrmi a0, 1
-; RV64D-NEXT:    vfcvt.f.xu.v v9, v9, v0.t
+; RV64D-NEXT:    vfcvt.f.xu.v v9, v9
 ; RV64D-NEXT:    li a1, 52
 ; RV64D-NEXT:    vsrl.vx v9, v9, a1
 ; RV64D-NEXT:    li a1, 1023
@@ -1614,10 +1604,9 @@ define <vscale x 2 x i64> @cttz_nxv2i64(<vscale x 2 x i64> %va) {
 ; RV32F-NEXT:    vsetvli a0, zero, e64, m2, ta, ma
 ; RV32F-NEXT:    vrsub.vi v10, v8, 0
 ; RV32F-NEXT:    vand.vv v10, v8, v10
-; RV32F-NEXT:    vmset.m v0
 ; RV32F-NEXT:    fsrmi a0, 1
 ; RV32F-NEXT:    vsetvli zero, zero, e32, m1, ta, ma
-; RV32F-NEXT:    vfncvt.f.xu.w v12, v10, v0.t
+; RV32F-NEXT:    vfncvt.f.xu.w v12, v10
 ; RV32F-NEXT:    vsrl.vi v10, v12, 23
 ; RV32F-NEXT:    vsetvli zero, zero, e64, m2, ta, ma
 ; RV32F-NEXT:    vzext.vf2 v12, v10
@@ -1634,10 +1623,9 @@ define <vscale x 2 x i64> @cttz_nxv2i64(<vscale x 2 x i64> %va) {
 ; RV64F-NEXT:    vsetvli a0, zero, e64, m2, ta, ma
 ; RV64F-NEXT:    vrsub.vi v10, v8, 0
 ; RV64F-NEXT:    vand.vv v10, v8, v10
-; RV64F-NEXT:    vmset.m v0
 ; RV64F-NEXT:    fsrmi a0, 1
 ; RV64F-NEXT:    vsetvli zero, zero, e32, m1, ta, ma
-; RV64F-NEXT:    vfncvt.f.xu.w v12, v10, v0.t
+; RV64F-NEXT:    vfncvt.f.xu.w v12, v10
 ; RV64F-NEXT:    vsrl.vi v10, v12, 23
 ; RV64F-NEXT:    vsetvli zero, zero, e64, m2, ta, ma
 ; RV64F-NEXT:    vzext.vf2 v12, v10
@@ -1654,9 +1642,8 @@ define <vscale x 2 x i64> @cttz_nxv2i64(<vscale x 2 x i64> %va) {
 ; RV32D-NEXT:    vsetvli a0, zero, e64, m2, ta, ma
 ; RV32D-NEXT:    vrsub.vi v10, v8, 0
 ; RV32D-NEXT:    vand.vv v10, v8, v10
-; RV32D-NEXT:    vmset.m v0
 ; RV32D-NEXT:    fsrmi a0, 1
-; RV32D-NEXT:    vfcvt.f.xu.v v10, v10, v0.t
+; RV32D-NEXT:    vfcvt.f.xu.v v10, v10
 ; RV32D-NEXT:    li a1, 52
 ; RV32D-NEXT:    vsrl.vx v10, v10, a1
 ; RV32D-NEXT:    li a1, 1023
@@ -1672,9 +1659,8 @@ define <vscale x 2 x i64> @cttz_nxv2i64(<vscale x 2 x i64> %va) {
 ; RV64D-NEXT:    vsetvli a0, zero, e64, m2, ta, ma
 ; RV64D-NEXT:    vrsub.vi v10, v8, 0
 ; RV64D-NEXT:    vand.vv v10, v8, v10
-; RV64D-NEXT:    vmset.m v0
 ; RV64D-NEXT:    fsrmi a0, 1
-; RV64D-NEXT:    vfcvt.f.xu.v v10, v10, v0.t
+; RV64D-NEXT:    vfcvt.f.xu.v v10, v10
 ; RV64D-NEXT:    li a1, 52
 ; RV64D-NEXT:    vsrl.vx v10, v10, a1
 ; RV64D-NEXT:    li a1, 1023
@@ -1782,10 +1768,9 @@ define <vscale x 4 x i64> @cttz_nxv4i64(<vscale x 4 x i64> %va) {
 ; RV32F-NEXT:    vsetvli a0, zero, e64, m4, ta, ma
 ; RV32F-NEXT:    vrsub.vi v12, v8, 0
 ; RV32F-NEXT:    vand.vv v12, v8, v12
-; RV32F-NEXT:    vmset.m v0
 ; RV32F-NEXT:    fsrmi a0, 1
 ; RV32F-NEXT:    vsetvli zero, zero, e32, m2, ta, ma
-; RV32F-NEXT:    vfncvt.f.xu.w v16, v12, v0.t
+; RV32F-NEXT:    vfncvt.f.xu.w v16, v12
 ; RV32F-NEXT:    vsrl.vi v12, v16, 23
 ; RV32F-NEXT:    vsetvli zero, zero, e64, m4, ta, ma
 ; RV32F-NEXT:    vzext.vf2 v16, v12
@@ -1802,10 +1787,9 @@ define <vscale x 4 x i64> @cttz_nxv4i64(<vscale x 4 x i64> %va) {
 ; RV64F-NEXT:    vsetvli a0, zero, e64, m4, ta, ma
 ; RV64F-NEXT:    vrsub.vi v12, v8, 0
 ; RV64F-NEXT:    vand.vv v12, v8, v12
-; RV64F-NEXT:    vmset.m v0
 ; RV64F-NEXT:    fsrmi a0, 1
 ; RV64F-NEXT:    vsetvli zero, zero, e32, m2, ta, ma
-; RV64F-NEXT:    vfncvt.f.xu.w v16, v12, v0.t
+; RV64F-NEXT:    vfncvt.f.xu.w v16, v12
 ; RV64F-NEXT:    vsrl.vi v12, v16, 23
 ; RV64F-NEXT:    vsetvli zero, zero, e64, m4, ta, ma
 ; RV64F-NEXT:    vzext.vf2 v16, v12
@@ -1822,9 +1806,8 @@ define <vscale x 4 x i64> @cttz_nxv4i64(<vscale x 4 x i64> %va) {
 ; RV32D-NEXT:    vsetvli a0, zero, e64, m4, ta, ma
 ; RV32D-NEXT:    vrsub.vi v12, v8, 0
 ; RV32D-NEXT:    vand.vv v12, v8, v12
-; RV32D-NEXT:    vmset.m v0
 ; RV32D-NEXT:    fsrmi a0, 1
-; RV32D-NEXT:    vfcvt.f.xu.v v12, v12, v0.t
+; RV32D-NEXT:    vfcvt.f.xu.v v12, v12
 ; RV32D-NEXT:    li a1, 52
 ; RV32D-NEXT:    vsrl.vx v12, v12, a1
 ; RV32D-NEXT:    li a1, 1023
@@ -1840,9 +1823,8 @@ define <vscale x 4 x i64> @cttz_nxv4i64(<vscale x 4 x i64> %va) {
 ; RV64D-NEXT:    vsetvli a0, zero, e64, m4, ta, ma
 ; RV64D-NEXT:    vrsub.vi v12, v8, 0
 ; RV64D-NEXT:    vand.vv v12, v8, v12
-; RV64D-NEXT:    vmset.m v0
 ; RV64D-NEXT:    fsrmi a0, 1
-; RV64D-NEXT:    vfcvt.f.xu.v v12, v12, v0.t
+; RV64D-NEXT:    vfcvt.f.xu.v v12, v12
 ; RV64D-NEXT:    li a1, 52
 ; RV64D-NEXT:    vsrl.vx v12, v12, a1
 ; RV64D-NEXT:    li a1, 1023
@@ -1950,10 +1932,9 @@ define <vscale x 8 x i64> @cttz_nxv8i64(<vscale x 8 x i64> %va) {
 ; RV32F-NEXT:    vsetvli a0, zero, e64, m8, ta, ma
 ; RV32F-NEXT:    vrsub.vi v16, v8, 0
 ; RV32F-NEXT:    vand.vv v16, v8, v16
-; RV32F-NEXT:    vmset.m v0
 ; RV32F-NEXT:    fsrmi a0, 1
 ; RV32F-NEXT:    vsetvli zero, zero, e32, m4, ta, ma
-; RV32F-NEXT:    vfncvt.f.xu.w v24, v16, v0.t
+; RV32F-NEXT:    vfncvt.f.xu.w v24, v16
 ; RV32F-NEXT:    vsrl.vi v16, v24, 23
 ; RV32F-NEXT:    vsetvli zero, zero, e64, m8, ta, ma
 ; RV32F-NEXT:    vzext.vf2 v24, v16
@@ -1970,10 +1951,9 @@ define <vscale x 8 x i64> @cttz_nxv8i64(<vscale x 8 x i64> %va) {
 ; RV64F-NEXT:    vsetvli a0, zero, e64, m8, ta, ma
 ; RV64F-NEXT:    vrsub.vi v16, v8, 0
 ; RV64F-NEXT:    vand.vv v16, v8, v16
-; RV64F-NEXT:    vmset.m v0
 ; RV64F-NEXT:    fsrmi a0, 1
 ; RV64F-NEXT:    vsetvli zero, zero, e32, m4, ta, ma
-; RV64F-NEXT:    vfncvt.f.xu.w v24, v16, v0.t
+; RV64F-NEXT:    vfncvt.f.xu.w v24, v16
 ; RV64F-NEXT:    vsrl.vi v16, v24, 23
 ; RV64F-NEXT:    vsetvli zero, zero, e64, m8, ta, ma
 ; RV64F-NEXT:    vzext.vf2 v24, v16
@@ -1990,9 +1970,8 @@ define <vscale x 8 x i64> @cttz_nxv8i64(<vscale x 8 x i64> %va) {
 ; RV32D-NEXT:    vsetvli a0, zero, e64, m8, ta, ma
 ; RV32D-NEXT:    vrsub.vi v16, v8, 0
 ; RV32D-NEXT:    vand.vv v16, v8, v16
-; RV32D-NEXT:    vmset.m v0
 ; RV32D-NEXT:    fsrmi a0, 1
-; RV32D-NEXT:    vfcvt.f.xu.v v16, v16, v0.t
+; RV32D-NEXT:    vfcvt.f.xu.v v16, v16
 ; RV32D-NEXT:    li a1, 52
 ; RV32D-NEXT:    vsrl.vx v16, v16, a1
 ; RV32D-NEXT:    li a1, 1023
@@ -2008,9 +1987,8 @@ define <vscale x 8 x i64> @cttz_nxv8i64(<vscale x 8 x i64> %va) {
 ; RV64D-NEXT:    vsetvli a0, zero, e64, m8, ta, ma
 ; RV64D-NEXT:    vrsub.vi v16, v8, 0
 ; RV64D-NEXT:    vand.vv v16, v8, v16
-; RV64D-NEXT:    vmset.m v0
 ; RV64D-NEXT:    fsrmi a0, 1
-; RV64D-NEXT:    vfcvt.f.xu.v v16, v16, v0.t
+; RV64D-NEXT:    vfcvt.f.xu.v v16, v16
 ; RV64D-NEXT:    li a1, 52
 ; RV64D-NEXT:    vsrl.vx v16, v16, a1
 ; RV64D-NEXT:    li a1, 1023
@@ -2884,9 +2862,8 @@ define <vscale x 1 x i32> @cttz_zero_undef_nxv1i32(<vscale x 1 x i32> %va) {
 ; CHECK-F-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
 ; CHECK-F-NEXT:    vrsub.vi v9, v8, 0
 ; CHECK-F-NEXT:    vand.vv v8, v8, v9
-; CHECK-F-NEXT:    vmset.m v0
 ; CHECK-F-NEXT:    fsrmi a0, 1
-; CHECK-F-NEXT:    vfcvt.f.xu.v v8, v8, v0.t
+; CHECK-F-NEXT:    vfcvt.f.xu.v v8, v8
 ; CHECK-F-NEXT:    vsrl.vi v8, v8, 23
 ; CHECK-F-NEXT:    li a1, 127
 ; CHECK-F-NEXT:    vsub.vx v8, v8, a1
@@ -2975,9 +2952,8 @@ define <vscale x 2 x i32> @cttz_zero_undef_nxv2i32(<vscale x 2 x i32> %va) {
 ; CHECK-F-NEXT:    vsetvli a0, zero, e32, m1, ta, ma
 ; CHECK-F-NEXT:    vrsub.vi v9, v8, 0
 ; CHECK-F-NEXT:    vand.vv v8, v8, v9
-; CHECK-F-NEXT:    vmset.m v0
 ; CHECK-F-NEXT:    fsrmi a0, 1
-; CHECK-F-NEXT:    vfcvt.f.xu.v v8, v8, v0.t
+; CHECK-F-NEXT:    vfcvt.f.xu.v v8, v8
 ; CHECK-F-NEXT:    vsrl.vi v8, v8, 23
 ; CHECK-F-NEXT:    li a1, 127
 ; CHECK-F-NEXT:    vsub.vx v8, v8, a1
@@ -3066,9 +3042,8 @@ define <vscale x 4 x i32> @cttz_zero_undef_nxv4i32(<vscale x 4 x i32> %va) {
 ; CHECK-F-NEXT:    vsetvli a0, zero, e32, m2, ta, ma
 ; CHECK-F-NEXT:    vrsub.vi v10, v8, 0
 ; CHECK-F-NEXT:    vand.vv v8, v8, v10
-; CHECK-F-NEXT:    vmset.m v0
 ; CHECK-F-NEXT:    fsrmi a0, 1
-; CHECK-F-NEXT:    vfcvt.f.xu.v v8, v8, v0.t
+; CHECK-F-NEXT:    vfcvt.f.xu.v v8, v8
 ; CHECK-F-NEXT:    vsrl.vi v8, v8, 23
 ; CHECK-F-NEXT:    li a1, 127
 ; CHECK-F-NEXT:    vsub.vx v8, v8, a1
@@ -3157,9 +3132,8 @@ define <vscale x 8 x i32> @cttz_zero_undef_nxv8i32(<vscale x 8 x i32> %va) {
 ; CHECK-F-NEXT:    vsetvli a0, zero, e32, m4, ta, ma
 ; CHECK-F-NEXT:    vrsub.vi v12, v8, 0
 ; CHECK-F-NEXT:    vand.vv v8, v8, v12
-; CHECK-F-NEXT:    vmset.m v0
 ; CHECK-F-NEXT:    fsrmi a0, 1
-; CHECK-F-NEXT:    vfcvt.f.xu.v v8, v8, v0.t
+; CHECK-F-NEXT:    vfcvt.f.xu.v v8, v8
 ; CHECK-F-NEXT:    vsrl.vi v8, v8, 23
 ; CHECK-F-NEXT:    li a1, 127
 ; CHECK-F-NEXT:    vsub.vx v8, v8, a1
@@ -3248,9 +3222,8 @@ define <vscale x 16 x i32> @cttz_zero_undef_nxv16i32(<vscale x 16 x i32> %va) {
 ; CHECK-F-NEXT:    vsetvli a0, zero, e32, m8, ta, ma
 ; CHECK-F-NEXT:    vrsub.vi v16, v8, 0
 ; CHECK-F-NEXT:    vand.vv v8, v8, v16
-; CHECK-F-NEXT:    vmset.m v0
 ; CHECK-F-NEXT:    fsrmi a0, 1
-; CHECK-F-NEXT:    vfcvt.f.xu.v v8, v8, v0.t
+; CHECK-F-NEXT:    vfcvt.f.xu.v v8, v8
 ; CHECK-F-NEXT:    vsrl.vi v8, v8, 23
 ; CHECK-F-NEXT:    li a1, 127
 ; CHECK-F-NEXT:    vsub.vx v8, v8, a1
@@ -3262,9 +3235,8 @@ define <vscale x 16 x i32> @cttz_zero_undef_nxv16i32(<vscale x 16 x i32> %va) {
 ; CHECK-D-NEXT:    vsetvli a0, zero, e32, m8, ta, ma
 ; CHECK-D-NEXT:    vrsub.vi v16, v8, 0
 ; CHECK-D-NEXT:    vand.vv v8, v8, v16
-; CHECK-D-NEXT:    vmset.m v0
 ; CHECK-D-NEXT:    fsrmi a0, 1
-; CHECK-D-NEXT:    vfcvt.f.xu.v v8, v8, v0.t
+; CHECK-D-NEXT:    vfcvt.f.xu.v v8, v8
 ; CHECK-D-NEXT:    vsrl.vi v8, v8, 23
 ; CHECK-D-NEXT:    li a1, 127
 ; CHECK-D-NEXT:    vsub.vx v8, v8, a1
@@ -3367,10 +3339,9 @@ define <vscale x 1 x i64> @cttz_zero_undef_nxv1i64(<vscale x 1 x i64> %va) {
 ; CHECK-F-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
 ; CHECK-F-NEXT:    vrsub.vi v9, v8, 0
 ; CHECK-F-NEXT:    vand.vv v8, v8, v9
-; CHECK-F-NEXT:    vmset.m v0
 ; CHECK-F-NEXT:    fsrmi a0, 1
 ; CHECK-F-NEXT:    vsetvli zero, zero, e32, mf2, ta, ma
-; CHECK-F-NEXT:    vfncvt.f.xu.w v9, v8, v0.t
+; CHECK-F-NEXT:    vfncvt.f.xu.w v9, v8
 ; CHECK-F-NEXT:    vsrl.vi v8, v9, 23
 ; CHECK-F-NEXT:    vsetvli zero, zero, e64, m1, ta, ma
 ; CHECK-F-NEXT:    vzext.vf2 v9, v8
@@ -3384,9 +3355,8 @@ define <vscale x 1 x i64> @cttz_zero_undef_nxv1i64(<vscale x 1 x i64> %va) {
 ; CHECK-D-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
 ; CHECK-D-NEXT:    vrsub.vi v9, v8, 0
 ; CHECK-D-NEXT:    vand.vv v8, v8, v9
-; CHECK-D-NEXT:    vmset.m v0
 ; CHECK-D-NEXT:    fsrmi a0, 1
-; CHECK-D-NEXT:    vfcvt.f.xu.v v8, v8, v0.t
+; CHECK-D-NEXT:    vfcvt.f.xu.v v8, v8
 ; CHECK-D-NEXT:    li a1, 52
 ; CHECK-D-NEXT:    vsrl.vx v8, v8, a1
 ; CHECK-D-NEXT:    li a1, 1023
@@ -3490,10 +3460,9 @@ define <vscale x 2 x i64> @cttz_zero_undef_nxv2i64(<vscale x 2 x i64> %va) {
 ; CHECK-F-NEXT:    vsetvli a0, zero, e64, m2, ta, ma
 ; CHECK-F-NEXT:    vrsub.vi v10, v8, 0
 ; CHECK-F-NEXT:    vand.vv v8, v8, v10
-; CHECK-F-NEXT:    vmset.m v0
 ; CHECK-F-NEXT:    fsrmi a0, 1
 ; CHECK-F-NEXT:    vsetvli zero, zero, e32, m1, ta, ma
-; CHECK-F-NEXT:    vfncvt.f.xu.w v10, v8, v0.t
+; CHECK-F-NEXT:    vfncvt.f.xu.w v10, v8
 ; CHECK-F-NEXT:    vsrl.vi v8, v10, 23
 ; CHECK-F-NEXT:    vsetvli zero, zero, e64, m2, ta, ma
 ; CHECK-F-NEXT:    vzext.vf2 v10, v8
@@ -3507,9 +3476,8 @@ define <vscale x 2 x i64> @cttz_zero_undef_nxv2i64(<vscale x 2 x i64> %va) {
 ; CHECK-D-NEXT:    vsetvli a0, zero, e64, m2, ta, ma
 ; CHECK-D-NEXT:    vrsub.vi v10, v8, 0
 ; CHECK-D-NEXT:    vand.vv v8, v8, v10
-; CHECK-D-NEXT:    vmset.m v0
 ; CHECK-D-NEXT:    fsrmi a0, 1
-; CHECK-D-NEXT:    vfcvt.f.xu.v v8, v8, v0.t
+; CHECK-D-NEXT:    vfcvt.f.xu.v v8, v8
 ; CHECK-D-NEXT:    li a1, 52
 ; CHECK-D-NEXT:    vsrl.vx v8, v8, a1
 ; CHECK-D-NEXT:    li a1, 1023
@@ -3613,10 +3581,9 @@ define <vscale x 4 x i64> @cttz_zero_undef_nxv4i64(<vscale x 4 x i64> %va) {
 ; CHECK-F-NEXT:    vsetvli a0, zero, e64, m4, ta, ma
 ; CHECK-F-NEXT:    vrsub.vi v12, v8, 0
 ; CHECK-F-NEXT:    vand.vv v8, v8, v12
-; CHECK-F-NEXT:    vmset.m v0
 ; CHECK-F-NEXT:    fsrmi a0, 1
 ; CHECK-F-NEXT:    vsetvli zero, zero, e32, m2, ta, ma
-; CHECK-F-NEXT:    vfncvt.f.xu.w v12, v8, v0.t
+; CHECK-F-NEXT:    vfncvt.f.xu.w v12, v8
 ; CHECK-F-NEXT:    vsrl.vi v8, v12, 23
 ; CHECK-F-NEXT:    vsetvli zero, zero, e64, m4, ta, ma
 ; CHECK-F-NEXT:    vzext.vf2 v12, v8
@@ -3630,9 +3597,8 @@ define <vscale x 4 x i64> @cttz_zero_undef_nxv4i64(<vscale x 4 x i64> %va) {
 ; CHECK-D-NEXT:    vsetvli a0, zero, e64, m4, ta, ma
 ; CHECK-D-NEXT:    vrsub.vi v12, v8, 0
 ; CHECK-D-NEXT:    vand.vv v8, v8, v12
-; CHECK-D-NEXT:    vmset.m v0
 ; CHECK-D-NEXT:    fsrmi a0, 1
-; CHECK-D-NEXT:    vfcvt.f.xu.v v8, v8, v0.t
+; CHECK-D-NEXT:    vfcvt.f.xu.v v8, v8
 ; CHECK-D-NEXT:    li a1, 52
 ; CHECK-D-NEXT:    vsrl.vx v8, v8, a1
 ; CHECK-D-NEXT:    li a1, 1023
@@ -3736,10 +3702,9 @@ define <vscale x 8 x i64> @cttz_zero_undef_nxv8i64(<vscale x 8 x i64> %va) {
 ; CHECK-F-NEXT:    vsetvli a0, zero, e64, m8, ta, ma
 ; CHECK-F-NEXT:    vrsub.vi v16, v8, 0
 ; CHECK-F-NEXT:    vand.vv v8, v8, v16
-; CHECK-F-NEXT:    vmset.m v0
 ; CHECK-F-NEXT:    fsrmi a0, 1
 ; CHECK-F-NEXT:    vsetvli zero, zero, e32, m4, ta, ma
-; CHECK-F-NEXT:    vfncvt.f.xu.w v16, v8, v0.t
+; CHECK-F-NEXT:    vfncvt.f.xu.w v16, v8
 ; CHECK-F-NEXT:    vsrl.vi v8, v16, 23
 ; CHECK-F-NEXT:    vsetvli zero, zero, e64, m8, ta, ma
 ; CHECK-F-NEXT:    vzext.vf2 v16, v8
@@ -3753,9 +3718,8 @@ define <vscale x 8 x i64> @cttz_zero_undef_nxv8i64(<vscale x 8 x i64> %va) {
 ; CHECK-D-NEXT:    vsetvli a0, zero, e64, m8, ta, ma
 ; CHECK-D-NEXT:    vrsub.vi v16, v8, 0
 ; CHECK-D-NEXT:    vand.vv v8, v8, v16
-; CHECK-D-NEXT:    vmset.m v0
 ; CHECK-D-NEXT:    fsrmi a0, 1
-; CHECK-D-NEXT:    vfcvt.f.xu.v v8, v8, v0.t
+; CHECK-D-NEXT:    vfcvt.f.xu.v v8, v8
 ; CHECK-D-NEXT:    li a1, 52
 ; CHECK-D-NEXT:    vsrl.vx v8, v8, a1
 ; CHECK-D-NEXT:    li a1, 1023

diff --git a/llvm/test/CodeGen/RISCV/rvv/double-round-conv.ll b/llvm/test/CodeGen/RISCV/rvv/double-round-conv.ll
index fd04a0f..2639cea 100644 (file)
--- a/llvm/test/CodeGen/RISCV/rvv/double-round-conv.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/double-round-conv.ll
@@ -677,20 +677,18 @@ define <vscale x 1 x i16> @ceil_nxv1f64_to_ui16(<vscale x 1 x double> %x) {
 define <vscale x 1 x i32> @ceil_nxv1f64_to_si32(<vscale x 1 x double> %x) {
 ; RV32-LABEL: ceil_nxv1f64_to_si32:
 ; RV32:       # %bb.0:
-; RV32-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
-; RV32-NEXT:    vmset.m v0
 ; RV32-NEXT:    fsrmi a0, 3
-; RV32-NEXT:    vfncvt.x.f.w v9, v8, v0.t
+; RV32-NEXT:    vsetvli a1, zero, e32, mf2, ta, ma
+; RV32-NEXT:    vfncvt.x.f.w v9, v8
 ; RV32-NEXT:    fsrm a0
 ; RV32-NEXT:    vmv1r.v v8, v9
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: ceil_nxv1f64_to_si32:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
-; RV64-NEXT:    vmset.m v0
 ; RV64-NEXT:    fsrmi a0, 3
-; RV64-NEXT:    vfncvt.x.f.w v9, v8, v0.t
+; RV64-NEXT:    vsetvli a1, zero, e32, mf2, ta, ma
+; RV64-NEXT:    vfncvt.x.f.w v9, v8
 ; RV64-NEXT:    fsrm a0
 ; RV64-NEXT:    vmv1r.v v8, v9
 ; RV64-NEXT:    ret
@@ -702,20 +700,18 @@ define <vscale x 1 x i32> @ceil_nxv1f64_to_si32(<vscale x 1 x double> %x) {
 define <vscale x 1 x i32> @ceil_nxv1f64_to_ui32(<vscale x 1 x double> %x) {
 ; RV32-LABEL: ceil_nxv1f64_to_ui32:
 ; RV32:       # %bb.0:
-; RV32-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
-; RV32-NEXT:    vmset.m v0
 ; RV32-NEXT:    fsrmi a0, 3
-; RV32-NEXT:    vfncvt.xu.f.w v9, v8, v0.t
+; RV32-NEXT:    vsetvli a1, zero, e32, mf2, ta, ma
+; RV32-NEXT:    vfncvt.xu.f.w v9, v8
 ; RV32-NEXT:    fsrm a0
 ; RV32-NEXT:    vmv1r.v v8, v9
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: ceil_nxv1f64_to_ui32:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
-; RV64-NEXT:    vmset.m v0
 ; RV64-NEXT:    fsrmi a0, 3
-; RV64-NEXT:    vfncvt.xu.f.w v9, v8, v0.t
+; RV64-NEXT:    vsetvli a1, zero, e32, mf2, ta, ma
+; RV64-NEXT:    vfncvt.xu.f.w v9, v8
 ; RV64-NEXT:    fsrm a0
 ; RV64-NEXT:    vmv1r.v v8, v9
 ; RV64-NEXT:    ret
@@ -727,19 +723,17 @@ define <vscale x 1 x i32> @ceil_nxv1f64_to_ui32(<vscale x 1 x double> %x) {
 define <vscale x 1 x i64> @ceil_nxv1f64_to_si64(<vscale x 1 x double> %x) {
 ; RV32-LABEL: ceil_nxv1f64_to_si64:
 ; RV32:       # %bb.0:
-; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
-; RV32-NEXT:    vmset.m v0
 ; RV32-NEXT:    fsrmi a0, 3
-; RV32-NEXT:    vfcvt.x.f.v v8, v8, v0.t
+; RV32-NEXT:    vsetvli a1, zero, e64, m1, ta, ma
+; RV32-NEXT:    vfcvt.x.f.v v8, v8
 ; RV32-NEXT:    fsrm a0
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: ceil_nxv1f64_to_si64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
-; RV64-NEXT:    vmset.m v0
 ; RV64-NEXT:    fsrmi a0, 3
-; RV64-NEXT:    vfcvt.x.f.v v8, v8, v0.t
+; RV64-NEXT:    vsetvli a1, zero, e64, m1, ta, ma
+; RV64-NEXT:    vfcvt.x.f.v v8, v8
 ; RV64-NEXT:    fsrm a0
 ; RV64-NEXT:    ret
   %a = call <vscale x 1 x double> @llvm.ceil.nxv1f64(<vscale x 1 x double> %x)
@@ -750,19 +744,17 @@ define <vscale x 1 x i64> @ceil_nxv1f64_to_si64(<vscale x 1 x double> %x) {
 define <vscale x 1 x i64> @ceil_nxv1f64_to_ui64(<vscale x 1 x double> %x) {
 ; RV32-LABEL: ceil_nxv1f64_to_ui64:
 ; RV32:       # %bb.0:
-; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
-; RV32-NEXT:    vmset.m v0
 ; RV32-NEXT:    fsrmi a0, 3
-; RV32-NEXT:    vfcvt.xu.f.v v8, v8, v0.t
+; RV32-NEXT:    vsetvli a1, zero, e64, m1, ta, ma
+; RV32-NEXT:    vfcvt.xu.f.v v8, v8
 ; RV32-NEXT:    fsrm a0
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: ceil_nxv1f64_to_ui64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
-; RV64-NEXT:    vmset.m v0
 ; RV64-NEXT:    fsrmi a0, 3
-; RV64-NEXT:    vfcvt.xu.f.v v8, v8, v0.t
+; RV64-NEXT:    vsetvli a1, zero, e64, m1, ta, ma
+; RV64-NEXT:    vfcvt.xu.f.v v8, v8
 ; RV64-NEXT:    fsrm a0
 ; RV64-NEXT:    ret
   %a = call <vscale x 1 x double> @llvm.ceil.nxv1f64(<vscale x 1 x double> %x)
@@ -959,20 +951,18 @@ define <vscale x 4 x i16> @ceil_nxv4f64_to_ui16(<vscale x 4 x double> %x) {
 define <vscale x 4 x i32> @ceil_nxv4f64_to_si32(<vscale x 4 x double> %x) {
 ; RV32-LABEL: ceil_nxv4f64_to_si32:
 ; RV32:       # %bb.0:
-; RV32-NEXT:    vsetvli a0, zero, e32, m2, ta, ma
-; RV32-NEXT:    vmset.m v0
 ; RV32-NEXT:    fsrmi a0, 3
-; RV32-NEXT:    vfncvt.x.f.w v12, v8, v0.t
+; RV32-NEXT:    vsetvli a1, zero, e32, m2, ta, ma
+; RV32-NEXT:    vfncvt.x.f.w v12, v8
 ; RV32-NEXT:    fsrm a0
 ; RV32-NEXT:    vmv.v.v v8, v12
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: ceil_nxv4f64_to_si32:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a0, zero, e32, m2, ta, ma
-; RV64-NEXT:    vmset.m v0
 ; RV64-NEXT:    fsrmi a0, 3
-; RV64-NEXT:    vfncvt.x.f.w v12, v8, v0.t
+; RV64-NEXT:    vsetvli a1, zero, e32, m2, ta, ma
+; RV64-NEXT:    vfncvt.x.f.w v12, v8
 ; RV64-NEXT:    fsrm a0
 ; RV64-NEXT:    vmv.v.v v8, v12
 ; RV64-NEXT:    ret
@@ -984,20 +974,18 @@ define <vscale x 4 x i32> @ceil_nxv4f64_to_si32(<vscale x 4 x double> %x) {
 define <vscale x 4 x i32> @ceil_nxv4f64_to_ui32(<vscale x 4 x double> %x) {
 ; RV32-LABEL: ceil_nxv4f64_to_ui32:
 ; RV32:       # %bb.0:
-; RV32-NEXT:    vsetvli a0, zero, e32, m2, ta, ma
-; RV32-NEXT:    vmset.m v0
 ; RV32-NEXT:    fsrmi a0, 3
-; RV32-NEXT:    vfncvt.xu.f.w v12, v8, v0.t
+; RV32-NEXT:    vsetvli a1, zero, e32, m2, ta, ma
+; RV32-NEXT:    vfncvt.xu.f.w v12, v8
 ; RV32-NEXT:    fsrm a0
 ; RV32-NEXT:    vmv.v.v v8, v12
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: ceil_nxv4f64_to_ui32:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a0, zero, e32, m2, ta, ma
-; RV64-NEXT:    vmset.m v0
 ; RV64-NEXT:    fsrmi a0, 3
-; RV64-NEXT:    vfncvt.xu.f.w v12, v8, v0.t
+; RV64-NEXT:    vsetvli a1, zero, e32, m2, ta, ma
+; RV64-NEXT:    vfncvt.xu.f.w v12, v8
 ; RV64-NEXT:    fsrm a0
 ; RV64-NEXT:    vmv.v.v v8, v12
 ; RV64-NEXT:    ret
@@ -1009,19 +997,17 @@ define <vscale x 4 x i32> @ceil_nxv4f64_to_ui32(<vscale x 4 x double> %x) {
 define <vscale x 4 x i64> @ceil_nxv4f64_to_si64(<vscale x 4 x double> %x) {
 ; RV32-LABEL: ceil_nxv4f64_to_si64:
 ; RV32:       # %bb.0:
-; RV32-NEXT:    vsetvli a0, zero, e64, m4, ta, ma
-; RV32-NEXT:    vmset.m v0
 ; RV32-NEXT:    fsrmi a0, 3
-; RV32-NEXT:    vfcvt.x.f.v v8, v8, v0.t
+; RV32-NEXT:    vsetvli a1, zero, e64, m4, ta, ma
+; RV32-NEXT:    vfcvt.x.f.v v8, v8
 ; RV32-NEXT:    fsrm a0
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: ceil_nxv4f64_to_si64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a0, zero, e64, m4, ta, ma
-; RV64-NEXT:    vmset.m v0
 ; RV64-NEXT:    fsrmi a0, 3
-; RV64-NEXT:    vfcvt.x.f.v v8, v8, v0.t
+; RV64-NEXT:    vsetvli a1, zero, e64, m4, ta, ma
+; RV64-NEXT:    vfcvt.x.f.v v8, v8
 ; RV64-NEXT:    fsrm a0
 ; RV64-NEXT:    ret
   %a = call <vscale x 4 x double> @llvm.ceil.nxv4f64(<vscale x 4 x double> %x)
@@ -1032,19 +1018,17 @@ define <vscale x 4 x i64> @ceil_nxv4f64_to_si64(<vscale x 4 x double> %x) {
 define <vscale x 4 x i64> @ceil_nxv4f64_to_ui64(<vscale x 4 x double> %x) {
 ; RV32-LABEL: ceil_nxv4f64_to_ui64:
 ; RV32:       # %bb.0:
-; RV32-NEXT:    vsetvli a0, zero, e64, m4, ta, ma
-; RV32-NEXT:    vmset.m v0
 ; RV32-NEXT:    fsrmi a0, 3
-; RV32-NEXT:    vfcvt.xu.f.v v8, v8, v0.t
+; RV32-NEXT:    vsetvli a1, zero, e64, m4, ta, ma
+; RV32-NEXT:    vfcvt.xu.f.v v8, v8
 ; RV32-NEXT:    fsrm a0
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: ceil_nxv4f64_to_ui64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a0, zero, e64, m4, ta, ma
-; RV64-NEXT:    vmset.m v0
 ; RV64-NEXT:    fsrmi a0, 3
-; RV64-NEXT:    vfcvt.xu.f.v v8, v8, v0.t
+; RV64-NEXT:    vsetvli a1, zero, e64, m4, ta, ma
+; RV64-NEXT:    vfcvt.xu.f.v v8, v8
 ; RV64-NEXT:    fsrm a0
 ; RV64-NEXT:    ret
   %a = call <vscale x 4 x double> @llvm.ceil.nxv4f64(<vscale x 4 x double> %x)

diff --git a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-ctlz.ll b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-ctlz.ll
index c7c959c..7de1fb2 100644 (file)
--- a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-ctlz.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-ctlz.ll
@@ -356,9 +356,8 @@ define void @ctlz_v4i32(ptr %x, ptr %y) nounwind {
 ; LMULMAX2-RV32F:       # %bb.0:
 ; LMULMAX2-RV32F-NEXT:    vsetivli zero, 4, e32, m1, ta, ma
 ; LMULMAX2-RV32F-NEXT:    vle32.v v8, (a0)
-; LMULMAX2-RV32F-NEXT:    vmset.m v0
 ; LMULMAX2-RV32F-NEXT:    fsrmi a1, 1
-; LMULMAX2-RV32F-NEXT:    vfcvt.f.xu.v v8, v8, v0.t
+; LMULMAX2-RV32F-NEXT:    vfcvt.f.xu.v v8, v8
 ; LMULMAX2-RV32F-NEXT:    fsrm a1
 ; LMULMAX2-RV32F-NEXT:    vsrl.vi v8, v8, 23
 ; LMULMAX2-RV32F-NEXT:    li a1, 158
@@ -372,9 +371,8 @@ define void @ctlz_v4i32(ptr %x, ptr %y) nounwind {
 ; LMULMAX2-RV64F:       # %bb.0:
 ; LMULMAX2-RV64F-NEXT:    vsetivli zero, 4, e32, m1, ta, ma
 ; LMULMAX2-RV64F-NEXT:    vle32.v v8, (a0)
-; LMULMAX2-RV64F-NEXT:    vmset.m v0
 ; LMULMAX2-RV64F-NEXT:    fsrmi a1, 1
-; LMULMAX2-RV64F-NEXT:    vfcvt.f.xu.v v8, v8, v0.t
+; LMULMAX2-RV64F-NEXT:    vfcvt.f.xu.v v8, v8
 ; LMULMAX2-RV64F-NEXT:    fsrm a1
 ; LMULMAX2-RV64F-NEXT:    vsrl.vi v8, v8, 23
 ; LMULMAX2-RV64F-NEXT:    li a1, 158
@@ -543,17 +541,15 @@ define void @ctlz_v2i64(ptr %x, ptr %y) nounwind {
 ;
 ; LMULMAX2-RV32F-LABEL: ctlz_v2i64:
 ; LMULMAX2-RV32F:       # %bb.0:
-; LMULMAX2-RV32F-NEXT:    vsetivli zero, 2, e32, mf2, ta, ma
+; LMULMAX2-RV32F-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
 ; LMULMAX2-RV32F-NEXT:    vle64.v v8, (a0)
-; LMULMAX2-RV32F-NEXT:    vmset.m v0
-; LMULMAX2-RV32F-NEXT:    fsrmi a1, 1
-; LMULMAX2-RV32F-NEXT:    vfncvt.f.xu.w v9, v8, v0.t
-; LMULMAX2-RV32F-NEXT:    fsrm a1
-; LMULMAX2-RV32F-NEXT:    vsrl.vi v8, v9, 23
 ; LMULMAX2-RV32F-NEXT:    li a1, 190
-; LMULMAX2-RV32F-NEXT:    vsetvli zero, zero, e64, m1, ta, ma
 ; LMULMAX2-RV32F-NEXT:    vmv.v.x v9, a1
+; LMULMAX2-RV32F-NEXT:    fsrmi a1, 1
 ; LMULMAX2-RV32F-NEXT:    vsetvli zero, zero, e32, mf2, ta, ma
+; LMULMAX2-RV32F-NEXT:    vfncvt.f.xu.w v10, v8
+; LMULMAX2-RV32F-NEXT:    fsrm a1
+; LMULMAX2-RV32F-NEXT:    vsrl.vi v8, v10, 23
 ; LMULMAX2-RV32F-NEXT:    vwsubu.wv v9, v9, v8
 ; LMULMAX2-RV32F-NEXT:    li a1, 64
 ; LMULMAX2-RV32F-NEXT:    vsetvli zero, zero, e64, m1, ta, ma
@@ -565,13 +561,12 @@ define void @ctlz_v2i64(ptr %x, ptr %y) nounwind {
 ; LMULMAX2-RV64F:       # %bb.0:
 ; LMULMAX2-RV64F-NEXT:    vsetivli zero, 2, e32, mf2, ta, ma
 ; LMULMAX2-RV64F-NEXT:    vle64.v v8, (a0)
-; LMULMAX2-RV64F-NEXT:    vmset.m v0
-; LMULMAX2-RV64F-NEXT:    fsrmi a1, 1
-; LMULMAX2-RV64F-NEXT:    vfncvt.f.xu.w v9, v8, v0.t
-; LMULMAX2-RV64F-NEXT:    fsrm a1
-; LMULMAX2-RV64F-NEXT:    vsrl.vi v8, v9, 23
 ; LMULMAX2-RV64F-NEXT:    li a1, 190
 ; LMULMAX2-RV64F-NEXT:    vmv.v.x v9, a1
+; LMULMAX2-RV64F-NEXT:    fsrmi a1, 1
+; LMULMAX2-RV64F-NEXT:    vfncvt.f.xu.w v10, v8
+; LMULMAX2-RV64F-NEXT:    fsrm a1
+; LMULMAX2-RV64F-NEXT:    vsrl.vi v8, v10, 23
 ; LMULMAX2-RV64F-NEXT:    vwsubu.vv v10, v9, v8
 ; LMULMAX2-RV64F-NEXT:    li a1, 64
 ; LMULMAX2-RV64F-NEXT:    vsetvli zero, zero, e64, m1, ta, ma
@@ -583,9 +578,8 @@ define void @ctlz_v2i64(ptr %x, ptr %y) nounwind {
 ; LMULMAX2-RV32D:       # %bb.0:
 ; LMULMAX2-RV32D-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
 ; LMULMAX2-RV32D-NEXT:    vle64.v v8, (a0)
-; LMULMAX2-RV32D-NEXT:    vmset.m v0
 ; LMULMAX2-RV32D-NEXT:    fsrmi a1, 1
-; LMULMAX2-RV32D-NEXT:    vfcvt.f.xu.v v8, v8, v0.t
+; LMULMAX2-RV32D-NEXT:    vfcvt.f.xu.v v8, v8
 ; LMULMAX2-RV32D-NEXT:    fsrm a1
 ; LMULMAX2-RV32D-NEXT:    li a1, 52
 ; LMULMAX2-RV32D-NEXT:    vsrl.vx v8, v8, a1
@@ -600,9 +594,8 @@ define void @ctlz_v2i64(ptr %x, ptr %y) nounwind {
 ; LMULMAX2-RV64D:       # %bb.0:
 ; LMULMAX2-RV64D-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
 ; LMULMAX2-RV64D-NEXT:    vle64.v v8, (a0)
-; LMULMAX2-RV64D-NEXT:    vmset.m v0
 ; LMULMAX2-RV64D-NEXT:    fsrmi a1, 1
-; LMULMAX2-RV64D-NEXT:    vfcvt.f.xu.v v8, v8, v0.t
+; LMULMAX2-RV64D-NEXT:    vfcvt.f.xu.v v8, v8
 ; LMULMAX2-RV64D-NEXT:    fsrm a1
 ; LMULMAX2-RV64D-NEXT:    li a1, 52
 ; LMULMAX2-RV64D-NEXT:    vsrl.vx v8, v8, a1
@@ -617,9 +610,8 @@ define void @ctlz_v2i64(ptr %x, ptr %y) nounwind {
 ; LMULMAX8:       # %bb.0:
 ; LMULMAX8-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
 ; LMULMAX8-NEXT:    vle64.v v8, (a0)
-; LMULMAX8-NEXT:    vmset.m v0
 ; LMULMAX8-NEXT:    fsrmi a1, 1
-; LMULMAX8-NEXT:    vfcvt.f.xu.v v8, v8, v0.t
+; LMULMAX8-NEXT:    vfcvt.f.xu.v v8, v8
 ; LMULMAX8-NEXT:    fsrm a1
 ; LMULMAX8-NEXT:    li a1, 52
 ; LMULMAX8-NEXT:    vsrl.vx v8, v8, a1
@@ -1025,9 +1017,8 @@ define void @ctlz_v8i32(ptr %x, ptr %y) nounwind {
 ; LMULMAX2-RV32F:       # %bb.0:
 ; LMULMAX2-RV32F-NEXT:    vsetivli zero, 8, e32, m2, ta, ma
 ; LMULMAX2-RV32F-NEXT:    vle32.v v8, (a0)
-; LMULMAX2-RV32F-NEXT:    vmset.m v0
 ; LMULMAX2-RV32F-NEXT:    fsrmi a1, 1
-; LMULMAX2-RV32F-NEXT:    vfcvt.f.xu.v v8, v8, v0.t
+; LMULMAX2-RV32F-NEXT:    vfcvt.f.xu.v v8, v8
 ; LMULMAX2-RV32F-NEXT:    fsrm a1
 ; LMULMAX2-RV32F-NEXT:    vsrl.vi v8, v8, 23
 ; LMULMAX2-RV32F-NEXT:    li a1, 158
@@ -1041,9 +1032,8 @@ define void @ctlz_v8i32(ptr %x, ptr %y) nounwind {
 ; LMULMAX2-RV64F:       # %bb.0:
 ; LMULMAX2-RV64F-NEXT:    vsetivli zero, 8, e32, m2, ta, ma
 ; LMULMAX2-RV64F-NEXT:    vle32.v v8, (a0)
-; LMULMAX2-RV64F-NEXT:    vmset.m v0
 ; LMULMAX2-RV64F-NEXT:    fsrmi a1, 1
-; LMULMAX2-RV64F-NEXT:    vfcvt.f.xu.v v8, v8, v0.t
+; LMULMAX2-RV64F-NEXT:    vfcvt.f.xu.v v8, v8
 ; LMULMAX2-RV64F-NEXT:    fsrm a1
 ; LMULMAX2-RV64F-NEXT:    vsrl.vi v8, v8, 23
 ; LMULMAX2-RV64F-NEXT:    li a1, 158
@@ -1057,9 +1047,8 @@ define void @ctlz_v8i32(ptr %x, ptr %y) nounwind {
 ; LMULMAX2-RV32D:       # %bb.0:
 ; LMULMAX2-RV32D-NEXT:    vsetivli zero, 8, e32, m2, ta, ma
 ; LMULMAX2-RV32D-NEXT:    vle32.v v8, (a0)
-; LMULMAX2-RV32D-NEXT:    vmset.m v0
 ; LMULMAX2-RV32D-NEXT:    fsrmi a1, 1
-; LMULMAX2-RV32D-NEXT:    vfcvt.f.xu.v v8, v8, v0.t
+; LMULMAX2-RV32D-NEXT:    vfcvt.f.xu.v v8, v8
 ; LMULMAX2-RV32D-NEXT:    fsrm a1
 ; LMULMAX2-RV32D-NEXT:    vsrl.vi v8, v8, 23
 ; LMULMAX2-RV32D-NEXT:    li a1, 158
@@ -1073,9 +1062,8 @@ define void @ctlz_v8i32(ptr %x, ptr %y) nounwind {
 ; LMULMAX2-RV64D:       # %bb.0:
 ; LMULMAX2-RV64D-NEXT:    vsetivli zero, 8, e32, m2, ta, ma
 ; LMULMAX2-RV64D-NEXT:    vle32.v v8, (a0)
-; LMULMAX2-RV64D-NEXT:    vmset.m v0
 ; LMULMAX2-RV64D-NEXT:    fsrmi a1, 1
-; LMULMAX2-RV64D-NEXT:    vfcvt.f.xu.v v8, v8, v0.t
+; LMULMAX2-RV64D-NEXT:    vfcvt.f.xu.v v8, v8
 ; LMULMAX2-RV64D-NEXT:    fsrm a1
 ; LMULMAX2-RV64D-NEXT:    vsrl.vi v8, v8, 23
 ; LMULMAX2-RV64D-NEXT:    li a1, 158
@@ -1216,17 +1204,15 @@ define void @ctlz_v4i64(ptr %x, ptr %y) nounwind {
 ;
 ; LMULMAX2-RV32F-LABEL: ctlz_v4i64:
 ; LMULMAX2-RV32F:       # %bb.0:
-; LMULMAX2-RV32F-NEXT:    vsetivli zero, 4, e32, m1, ta, ma
+; LMULMAX2-RV32F-NEXT:    vsetivli zero, 4, e64, m2, ta, ma
 ; LMULMAX2-RV32F-NEXT:    vle64.v v8, (a0)
-; LMULMAX2-RV32F-NEXT:    vmset.m v0
-; LMULMAX2-RV32F-NEXT:    fsrmi a1, 1
-; LMULMAX2-RV32F-NEXT:    vfncvt.f.xu.w v10, v8, v0.t
-; LMULMAX2-RV32F-NEXT:    fsrm a1
-; LMULMAX2-RV32F-NEXT:    vsrl.vi v8, v10, 23
 ; LMULMAX2-RV32F-NEXT:    li a1, 190
-; LMULMAX2-RV32F-NEXT:    vsetvli zero, zero, e64, m2, ta, ma
 ; LMULMAX2-RV32F-NEXT:    vmv.v.x v10, a1
+; LMULMAX2-RV32F-NEXT:    fsrmi a1, 1
 ; LMULMAX2-RV32F-NEXT:    vsetvli zero, zero, e32, m1, ta, ma
+; LMULMAX2-RV32F-NEXT:    vfncvt.f.xu.w v12, v8
+; LMULMAX2-RV32F-NEXT:    fsrm a1
+; LMULMAX2-RV32F-NEXT:    vsrl.vi v8, v12, 23
 ; LMULMAX2-RV32F-NEXT:    vwsubu.wv v10, v10, v8
 ; LMULMAX2-RV32F-NEXT:    li a1, 64
 ; LMULMAX2-RV32F-NEXT:    vsetvli zero, zero, e64, m2, ta, ma
@@ -1238,17 +1224,16 @@ define void @ctlz_v4i64(ptr %x, ptr %y) nounwind {
 ; LMULMAX2-RV64F:       # %bb.0:
 ; LMULMAX2-RV64F-NEXT:    vsetivli zero, 4, e32, m1, ta, ma
 ; LMULMAX2-RV64F-NEXT:    vle64.v v8, (a0)
-; LMULMAX2-RV64F-NEXT:    vmset.m v0
+; LMULMAX2-RV64F-NEXT:    li a1, 190
+; LMULMAX2-RV64F-NEXT:    vmv.v.x v10, a1
 ; LMULMAX2-RV64F-NEXT:    fsrmi a1, 1
-; LMULMAX2-RV64F-NEXT:    vfncvt.f.xu.w v10, v8, v0.t
+; LMULMAX2-RV64F-NEXT:    vfncvt.f.xu.w v11, v8
 ; LMULMAX2-RV64F-NEXT:    fsrm a1
-; LMULMAX2-RV64F-NEXT:    vsrl.vi v8, v10, 23
-; LMULMAX2-RV64F-NEXT:    li a1, 190
-; LMULMAX2-RV64F-NEXT:    vmv.v.x v9, a1
-; LMULMAX2-RV64F-NEXT:    vwsubu.vv v10, v9, v8
+; LMULMAX2-RV64F-NEXT:    vsrl.vi v8, v11, 23
+; LMULMAX2-RV64F-NEXT:    vwsubu.vv v12, v10, v8
 ; LMULMAX2-RV64F-NEXT:    li a1, 64
 ; LMULMAX2-RV64F-NEXT:    vsetvli zero, zero, e64, m2, ta, ma
-; LMULMAX2-RV64F-NEXT:    vminu.vx v8, v10, a1
+; LMULMAX2-RV64F-NEXT:    vminu.vx v8, v12, a1
 ; LMULMAX2-RV64F-NEXT:    vse64.v v8, (a0)
 ; LMULMAX2-RV64F-NEXT:    ret
 ;
@@ -1256,9 +1241,8 @@ define void @ctlz_v4i64(ptr %x, ptr %y) nounwind {
 ; LMULMAX2-RV32D:       # %bb.0:
 ; LMULMAX2-RV32D-NEXT:    vsetivli zero, 4, e64, m2, ta, ma
 ; LMULMAX2-RV32D-NEXT:    vle64.v v8, (a0)
-; LMULMAX2-RV32D-NEXT:    vmset.m v0
 ; LMULMAX2-RV32D-NEXT:    fsrmi a1, 1
-; LMULMAX2-RV32D-NEXT:    vfcvt.f.xu.v v8, v8, v0.t
+; LMULMAX2-RV32D-NEXT:    vfcvt.f.xu.v v8, v8
 ; LMULMAX2-RV32D-NEXT:    fsrm a1
 ; LMULMAX2-RV32D-NEXT:    li a1, 52
 ; LMULMAX2-RV32D-NEXT:    vsrl.vx v8, v8, a1
@@ -1273,9 +1257,8 @@ define void @ctlz_v4i64(ptr %x, ptr %y) nounwind {
 ; LMULMAX2-RV64D:       # %bb.0:
 ; LMULMAX2-RV64D-NEXT:    vsetivli zero, 4, e64, m2, ta, ma
 ; LMULMAX2-RV64D-NEXT:    vle64.v v8, (a0)
-; LMULMAX2-RV64D-NEXT:    vmset.m v0
 ; LMULMAX2-RV64D-NEXT:    fsrmi a1, 1
-; LMULMAX2-RV64D-NEXT:    vfcvt.f.xu.v v8, v8, v0.t
+; LMULMAX2-RV64D-NEXT:    vfcvt.f.xu.v v8, v8
 ; LMULMAX2-RV64D-NEXT:    fsrm a1
 ; LMULMAX2-RV64D-NEXT:    li a1, 52
 ; LMULMAX2-RV64D-NEXT:    vsrl.vx v8, v8, a1
@@ -1290,9 +1273,8 @@ define void @ctlz_v4i64(ptr %x, ptr %y) nounwind {
 ; LMULMAX8:       # %bb.0:
 ; LMULMAX8-NEXT:    vsetivli zero, 4, e64, m2, ta, ma
 ; LMULMAX8-NEXT:    vle64.v v8, (a0)
-; LMULMAX8-NEXT:    vmset.m v0
 ; LMULMAX8-NEXT:    fsrmi a1, 1
-; LMULMAX8-NEXT:    vfcvt.f.xu.v v8, v8, v0.t
+; LMULMAX8-NEXT:    vfcvt.f.xu.v v8, v8
 ; LMULMAX8-NEXT:    fsrm a1
 ; LMULMAX8-NEXT:    li a1, 52
 ; LMULMAX8-NEXT:    vsrl.vx v8, v8, a1

diff --git a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-cttz.ll b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-cttz.ll
index f2407bb..f32f6d0 100644 (file)
--- a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-cttz.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-cttz.ll
@@ -339,9 +339,8 @@ define void @cttz_v4i32(ptr %x, ptr %y) nounwind {
 ; LMULMAX2-RV32F-NEXT:    vle32.v v8, (a0)
 ; LMULMAX2-RV32F-NEXT:    vrsub.vi v9, v8, 0
 ; LMULMAX2-RV32F-NEXT:    vand.vv v9, v8, v9
-; LMULMAX2-RV32F-NEXT:    vmset.m v0
 ; LMULMAX2-RV32F-NEXT:    fsrmi a1, 1
-; LMULMAX2-RV32F-NEXT:    vfcvt.f.xu.v v9, v9, v0.t
+; LMULMAX2-RV32F-NEXT:    vfcvt.f.xu.v v9, v9
 ; LMULMAX2-RV32F-NEXT:    fsrm a1
 ; LMULMAX2-RV32F-NEXT:    vsrl.vi v9, v9, 23
 ; LMULMAX2-RV32F-NEXT:    li a1, 127
@@ -358,9 +357,8 @@ define void @cttz_v4i32(ptr %x, ptr %y) nounwind {
 ; LMULMAX2-RV64F-NEXT:    vle32.v v8, (a0)
 ; LMULMAX2-RV64F-NEXT:    vrsub.vi v9, v8, 0
 ; LMULMAX2-RV64F-NEXT:    vand.vv v9, v8, v9
-; LMULMAX2-RV64F-NEXT:    vmset.m v0
 ; LMULMAX2-RV64F-NEXT:    fsrmi a1, 1
-; LMULMAX2-RV64F-NEXT:    vfcvt.f.xu.v v9, v9, v0.t
+; LMULMAX2-RV64F-NEXT:    vfcvt.f.xu.v v9, v9
 ; LMULMAX2-RV64F-NEXT:    fsrm a1
 ; LMULMAX2-RV64F-NEXT:    vsrl.vi v9, v9, 23
 ; LMULMAX2-RV64F-NEXT:    li a1, 127
@@ -520,26 +518,24 @@ define void @cttz_v2i64(ptr %x, ptr %y) nounwind {
 ; LMULMAX2-RV32F-LABEL: cttz_v2i64:
 ; LMULMAX2-RV32F:       # %bb.0:
 ; LMULMAX2-RV32F-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
-; LMULMAX2-RV32F-NEXT:    vle64.v v9, (a0)
+; LMULMAX2-RV32F-NEXT:    vle64.v v8, (a0)
 ; LMULMAX2-RV32F-NEXT:    vsetivli zero, 4, e32, m1, ta, ma
-; LMULMAX2-RV32F-NEXT:    vmv.v.i v10, 0
+; LMULMAX2-RV32F-NEXT:    vmv.v.i v9, 0
 ; LMULMAX2-RV32F-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
-; LMULMAX2-RV32F-NEXT:    vmseq.vv v8, v9, v10
-; LMULMAX2-RV32F-NEXT:    vsub.vv v10, v10, v9
-; LMULMAX2-RV32F-NEXT:    vand.vv v9, v9, v10
-; LMULMAX2-RV32F-NEXT:    vmset.m v0
+; LMULMAX2-RV32F-NEXT:    vmseq.vv v0, v8, v9
+; LMULMAX2-RV32F-NEXT:    vsub.vv v9, v9, v8
+; LMULMAX2-RV32F-NEXT:    vand.vv v8, v8, v9
 ; LMULMAX2-RV32F-NEXT:    fsrmi a1, 1
 ; LMULMAX2-RV32F-NEXT:    vsetvli zero, zero, e32, mf2, ta, ma
-; LMULMAX2-RV32F-NEXT:    vfncvt.f.xu.w v10, v9, v0.t
+; LMULMAX2-RV32F-NEXT:    vfncvt.f.xu.w v9, v8
 ; LMULMAX2-RV32F-NEXT:    fsrm a1
-; LMULMAX2-RV32F-NEXT:    vsrl.vi v9, v10, 23
+; LMULMAX2-RV32F-NEXT:    vsrl.vi v8, v9, 23
 ; LMULMAX2-RV32F-NEXT:    vsetvli zero, zero, e64, m1, ta, ma
-; LMULMAX2-RV32F-NEXT:    vzext.vf2 v10, v9
+; LMULMAX2-RV32F-NEXT:    vzext.vf2 v9, v8
 ; LMULMAX2-RV32F-NEXT:    li a1, 127
-; LMULMAX2-RV32F-NEXT:    vsub.vx v9, v10, a1
+; LMULMAX2-RV32F-NEXT:    vsub.vx v8, v9, a1
 ; LMULMAX2-RV32F-NEXT:    li a1, 64
-; LMULMAX2-RV32F-NEXT:    vmv.v.v v0, v8
-; LMULMAX2-RV32F-NEXT:    vmerge.vxm v8, v9, a1, v0
+; LMULMAX2-RV32F-NEXT:    vmerge.vxm v8, v8, a1, v0
 ; LMULMAX2-RV32F-NEXT:    vse64.v v8, (a0)
 ; LMULMAX2-RV32F-NEXT:    ret
 ;
@@ -549,10 +545,9 @@ define void @cttz_v2i64(ptr %x, ptr %y) nounwind {
 ; LMULMAX2-RV64F-NEXT:    vle64.v v8, (a0)
 ; LMULMAX2-RV64F-NEXT:    vrsub.vi v9, v8, 0
 ; LMULMAX2-RV64F-NEXT:    vand.vv v9, v8, v9
-; LMULMAX2-RV64F-NEXT:    vmset.m v0
 ; LMULMAX2-RV64F-NEXT:    fsrmi a1, 1
 ; LMULMAX2-RV64F-NEXT:    vsetvli zero, zero, e32, mf2, ta, ma
-; LMULMAX2-RV64F-NEXT:    vfncvt.f.xu.w v10, v9, v0.t
+; LMULMAX2-RV64F-NEXT:    vfncvt.f.xu.w v10, v9
 ; LMULMAX2-RV64F-NEXT:    fsrm a1
 ; LMULMAX2-RV64F-NEXT:    vsrl.vi v9, v10, 23
 ; LMULMAX2-RV64F-NEXT:    li a1, 127
@@ -567,24 +562,22 @@ define void @cttz_v2i64(ptr %x, ptr %y) nounwind {
 ; LMULMAX2-RV32D-LABEL: cttz_v2i64:
 ; LMULMAX2-RV32D:       # %bb.0:
 ; LMULMAX2-RV32D-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
-; LMULMAX2-RV32D-NEXT:    vle64.v v9, (a0)
+; LMULMAX2-RV32D-NEXT:    vle64.v v8, (a0)
 ; LMULMAX2-RV32D-NEXT:    vsetivli zero, 4, e32, m1, ta, ma
-; LMULMAX2-RV32D-NEXT:    vmv.v.i v10, 0
+; LMULMAX2-RV32D-NEXT:    vmv.v.i v9, 0
 ; LMULMAX2-RV32D-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
-; LMULMAX2-RV32D-NEXT:    vmseq.vv v8, v9, v10
-; LMULMAX2-RV32D-NEXT:    vsub.vv v10, v10, v9
-; LMULMAX2-RV32D-NEXT:    vand.vv v9, v9, v10
-; LMULMAX2-RV32D-NEXT:    vmset.m v0
+; LMULMAX2-RV32D-NEXT:    vmseq.vv v0, v8, v9
+; LMULMAX2-RV32D-NEXT:    vsub.vv v9, v9, v8
+; LMULMAX2-RV32D-NEXT:    vand.vv v8, v8, v9
 ; LMULMAX2-RV32D-NEXT:    fsrmi a1, 1
-; LMULMAX2-RV32D-NEXT:    vfcvt.f.xu.v v9, v9, v0.t
+; LMULMAX2-RV32D-NEXT:    vfcvt.f.xu.v v8, v8
 ; LMULMAX2-RV32D-NEXT:    fsrm a1
 ; LMULMAX2-RV32D-NEXT:    li a1, 52
-; LMULMAX2-RV32D-NEXT:    vsrl.vx v9, v9, a1
+; LMULMAX2-RV32D-NEXT:    vsrl.vx v8, v8, a1
 ; LMULMAX2-RV32D-NEXT:    li a1, 1023
-; LMULMAX2-RV32D-NEXT:    vsub.vx v9, v9, a1
+; LMULMAX2-RV32D-NEXT:    vsub.vx v8, v8, a1
 ; LMULMAX2-RV32D-NEXT:    li a1, 64
-; LMULMAX2-RV32D-NEXT:    vmv.v.v v0, v8
-; LMULMAX2-RV32D-NEXT:    vmerge.vxm v8, v9, a1, v0
+; LMULMAX2-RV32D-NEXT:    vmerge.vxm v8, v8, a1, v0
 ; LMULMAX2-RV32D-NEXT:    vse64.v v8, (a0)
 ; LMULMAX2-RV32D-NEXT:    ret
 ;
@@ -594,9 +587,8 @@ define void @cttz_v2i64(ptr %x, ptr %y) nounwind {
 ; LMULMAX2-RV64D-NEXT:    vle64.v v8, (a0)
 ; LMULMAX2-RV64D-NEXT:    vrsub.vi v9, v8, 0
 ; LMULMAX2-RV64D-NEXT:    vand.vv v9, v8, v9
-; LMULMAX2-RV64D-NEXT:    vmset.m v0
 ; LMULMAX2-RV64D-NEXT:    fsrmi a1, 1
-; LMULMAX2-RV64D-NEXT:    vfcvt.f.xu.v v9, v9, v0.t
+; LMULMAX2-RV64D-NEXT:    vfcvt.f.xu.v v9, v9
 ; LMULMAX2-RV64D-NEXT:    fsrm a1
 ; LMULMAX2-RV64D-NEXT:    li a1, 52
 ; LMULMAX2-RV64D-NEXT:    vsrl.vx v9, v9, a1
@@ -611,24 +603,22 @@ define void @cttz_v2i64(ptr %x, ptr %y) nounwind {
 ; LMULMAX8-RV32-LABEL: cttz_v2i64:
 ; LMULMAX8-RV32:       # %bb.0:
 ; LMULMAX8-RV32-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
-; LMULMAX8-RV32-NEXT:    vle64.v v9, (a0)
+; LMULMAX8-RV32-NEXT:    vle64.v v8, (a0)
 ; LMULMAX8-RV32-NEXT:    vsetivli zero, 4, e32, m1, ta, ma
-; LMULMAX8-RV32-NEXT:    vmv.v.i v10, 0
+; LMULMAX8-RV32-NEXT:    vmv.v.i v9, 0
 ; LMULMAX8-RV32-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
-; LMULMAX8-RV32-NEXT:    vmseq.vv v8, v9, v10
-; LMULMAX8-RV32-NEXT:    vsub.vv v10, v10, v9
-; LMULMAX8-RV32-NEXT:    vand.vv v9, v9, v10
-; LMULMAX8-RV32-NEXT:    vmset.m v0
+; LMULMAX8-RV32-NEXT:    vmseq.vv v0, v8, v9
+; LMULMAX8-RV32-NEXT:    vsub.vv v9, v9, v8
+; LMULMAX8-RV32-NEXT:    vand.vv v8, v8, v9
 ; LMULMAX8-RV32-NEXT:    fsrmi a1, 1
-; LMULMAX8-RV32-NEXT:    vfcvt.f.xu.v v9, v9, v0.t
+; LMULMAX8-RV32-NEXT:    vfcvt.f.xu.v v8, v8
 ; LMULMAX8-RV32-NEXT:    fsrm a1
 ; LMULMAX8-RV32-NEXT:    li a1, 52
-; LMULMAX8-RV32-NEXT:    vsrl.vx v9, v9, a1
+; LMULMAX8-RV32-NEXT:    vsrl.vx v8, v8, a1
 ; LMULMAX8-RV32-NEXT:    li a1, 1023
-; LMULMAX8-RV32-NEXT:    vsub.vx v9, v9, a1
+; LMULMAX8-RV32-NEXT:    vsub.vx v8, v8, a1
 ; LMULMAX8-RV32-NEXT:    li a1, 64
-; LMULMAX8-RV32-NEXT:    vmv.v.v v0, v8
-; LMULMAX8-RV32-NEXT:    vmerge.vxm v8, v9, a1, v0
+; LMULMAX8-RV32-NEXT:    vmerge.vxm v8, v8, a1, v0
 ; LMULMAX8-RV32-NEXT:    vse64.v v8, (a0)
 ; LMULMAX8-RV32-NEXT:    ret
 ;
@@ -638,9 +628,8 @@ define void @cttz_v2i64(ptr %x, ptr %y) nounwind {
 ; LMULMAX8-RV64-NEXT:    vle64.v v8, (a0)
 ; LMULMAX8-RV64-NEXT:    vrsub.vi v9, v8, 0
 ; LMULMAX8-RV64-NEXT:    vand.vv v9, v8, v9
-; LMULMAX8-RV64-NEXT:    vmset.m v0
 ; LMULMAX8-RV64-NEXT:    fsrmi a1, 1
-; LMULMAX8-RV64-NEXT:    vfcvt.f.xu.v v9, v9, v0.t
+; LMULMAX8-RV64-NEXT:    vfcvt.f.xu.v v9, v9
 ; LMULMAX8-RV64-NEXT:    fsrm a1
 ; LMULMAX8-RV64-NEXT:    li a1, 52
 ; LMULMAX8-RV64-NEXT:    vsrl.vx v9, v9, a1
@@ -999,9 +988,8 @@ define void @cttz_v8i32(ptr %x, ptr %y) nounwind {
 ; LMULMAX2-RV32F-NEXT:    vle32.v v8, (a0)
 ; LMULMAX2-RV32F-NEXT:    vrsub.vi v10, v8, 0
 ; LMULMAX2-RV32F-NEXT:    vand.vv v10, v8, v10
-; LMULMAX2-RV32F-NEXT:    vmset.m v0
 ; LMULMAX2-RV32F-NEXT:    fsrmi a1, 1
-; LMULMAX2-RV32F-NEXT:    vfcvt.f.xu.v v10, v10, v0.t
+; LMULMAX2-RV32F-NEXT:    vfcvt.f.xu.v v10, v10
 ; LMULMAX2-RV32F-NEXT:    fsrm a1
 ; LMULMAX2-RV32F-NEXT:    vsrl.vi v10, v10, 23
 ; LMULMAX2-RV32F-NEXT:    li a1, 127
@@ -1018,9 +1006,8 @@ define void @cttz_v8i32(ptr %x, ptr %y) nounwind {
 ; LMULMAX2-RV64F-NEXT:    vle32.v v8, (a0)
 ; LMULMAX2-RV64F-NEXT:    vrsub.vi v10, v8, 0
 ; LMULMAX2-RV64F-NEXT:    vand.vv v10, v8, v10
-; LMULMAX2-RV64F-NEXT:    vmset.m v0
 ; LMULMAX2-RV64F-NEXT:    fsrmi a1, 1
-; LMULMAX2-RV64F-NEXT:    vfcvt.f.xu.v v10, v10, v0.t
+; LMULMAX2-RV64F-NEXT:    vfcvt.f.xu.v v10, v10
 ; LMULMAX2-RV64F-NEXT:    fsrm a1
 ; LMULMAX2-RV64F-NEXT:    vsrl.vi v10, v10, 23
 ; LMULMAX2-RV64F-NEXT:    li a1, 127
@@ -1037,9 +1024,8 @@ define void @cttz_v8i32(ptr %x, ptr %y) nounwind {
 ; LMULMAX2-RV32D-NEXT:    vle32.v v8, (a0)
 ; LMULMAX2-RV32D-NEXT:    vrsub.vi v10, v8, 0
 ; LMULMAX2-RV32D-NEXT:    vand.vv v10, v8, v10
-; LMULMAX2-RV32D-NEXT:    vmset.m v0
 ; LMULMAX2-RV32D-NEXT:    fsrmi a1, 1
-; LMULMAX2-RV32D-NEXT:    vfcvt.f.xu.v v10, v10, v0.t
+; LMULMAX2-RV32D-NEXT:    vfcvt.f.xu.v v10, v10
 ; LMULMAX2-RV32D-NEXT:    fsrm a1
 ; LMULMAX2-RV32D-NEXT:    vsrl.vi v10, v10, 23
 ; LMULMAX2-RV32D-NEXT:    li a1, 127
@@ -1056,9 +1042,8 @@ define void @cttz_v8i32(ptr %x, ptr %y) nounwind {
 ; LMULMAX2-RV64D-NEXT:    vle32.v v8, (a0)
 ; LMULMAX2-RV64D-NEXT:    vrsub.vi v10, v8, 0
 ; LMULMAX2-RV64D-NEXT:    vand.vv v10, v8, v10
-; LMULMAX2-RV64D-NEXT:    vmset.m v0
 ; LMULMAX2-RV64D-NEXT:    fsrmi a1, 1
-; LMULMAX2-RV64D-NEXT:    vfcvt.f.xu.v v10, v10, v0.t
+; LMULMAX2-RV64D-NEXT:    vfcvt.f.xu.v v10, v10
 ; LMULMAX2-RV64D-NEXT:    fsrm a1
 ; LMULMAX2-RV64D-NEXT:    vsrl.vi v10, v10, 23
 ; LMULMAX2-RV64D-NEXT:    li a1, 127
@@ -1184,26 +1169,24 @@ define void @cttz_v4i64(ptr %x, ptr %y) nounwind {
 ; LMULMAX2-RV32F-LABEL: cttz_v4i64:
 ; LMULMAX2-RV32F:       # %bb.0:
 ; LMULMAX2-RV32F-NEXT:    vsetivli zero, 4, e64, m2, ta, ma
-; LMULMAX2-RV32F-NEXT:    vle64.v v10, (a0)
+; LMULMAX2-RV32F-NEXT:    vle64.v v8, (a0)
 ; LMULMAX2-RV32F-NEXT:    vsetivli zero, 8, e32, m2, ta, ma
-; LMULMAX2-RV32F-NEXT:    vmv.v.i v12, 0
+; LMULMAX2-RV32F-NEXT:    vmv.v.i v10, 0
 ; LMULMAX2-RV32F-NEXT:    vsetivli zero, 4, e64, m2, ta, ma
-; LMULMAX2-RV32F-NEXT:    vmseq.vv v8, v10, v12
-; LMULMAX2-RV32F-NEXT:    vsub.vv v12, v12, v10
-; LMULMAX2-RV32F-NEXT:    vand.vv v10, v10, v12
-; LMULMAX2-RV32F-NEXT:    vmset.m v0
+; LMULMAX2-RV32F-NEXT:    vmseq.vv v0, v8, v10
+; LMULMAX2-RV32F-NEXT:    vsub.vv v10, v10, v8
+; LMULMAX2-RV32F-NEXT:    vand.vv v8, v8, v10
 ; LMULMAX2-RV32F-NEXT:    fsrmi a1, 1
 ; LMULMAX2-RV32F-NEXT:    vsetvli zero, zero, e32, m1, ta, ma
-; LMULMAX2-RV32F-NEXT:    vfncvt.f.xu.w v9, v10, v0.t
+; LMULMAX2-RV32F-NEXT:    vfncvt.f.xu.w v10, v8
 ; LMULMAX2-RV32F-NEXT:    fsrm a1
-; LMULMAX2-RV32F-NEXT:    vsrl.vi v9, v9, 23
+; LMULMAX2-RV32F-NEXT:    vsrl.vi v8, v10, 23
 ; LMULMAX2-RV32F-NEXT:    vsetvli zero, zero, e64, m2, ta, ma
-; LMULMAX2-RV32F-NEXT:    vzext.vf2 v10, v9
+; LMULMAX2-RV32F-NEXT:    vzext.vf2 v10, v8
 ; LMULMAX2-RV32F-NEXT:    li a1, 127
-; LMULMAX2-RV32F-NEXT:    vsub.vx v10, v10, a1
+; LMULMAX2-RV32F-NEXT:    vsub.vx v8, v10, a1
 ; LMULMAX2-RV32F-NEXT:    li a1, 64
-; LMULMAX2-RV32F-NEXT:    vmv1r.v v0, v8
-; LMULMAX2-RV32F-NEXT:    vmerge.vxm v8, v10, a1, v0
+; LMULMAX2-RV32F-NEXT:    vmerge.vxm v8, v8, a1, v0
 ; LMULMAX2-RV32F-NEXT:    vse64.v v8, (a0)
 ; LMULMAX2-RV32F-NEXT:    ret
 ;
@@ -1213,10 +1196,9 @@ define void @cttz_v4i64(ptr %x, ptr %y) nounwind {
 ; LMULMAX2-RV64F-NEXT:    vle64.v v8, (a0)
 ; LMULMAX2-RV64F-NEXT:    vrsub.vi v10, v8, 0
 ; LMULMAX2-RV64F-NEXT:    vand.vv v10, v8, v10
-; LMULMAX2-RV64F-NEXT:    vmset.m v0
 ; LMULMAX2-RV64F-NEXT:    fsrmi a1, 1
 ; LMULMAX2-RV64F-NEXT:    vsetvli zero, zero, e32, m1, ta, ma
-; LMULMAX2-RV64F-NEXT:    vfncvt.f.xu.w v12, v10, v0.t
+; LMULMAX2-RV64F-NEXT:    vfncvt.f.xu.w v12, v10
 ; LMULMAX2-RV64F-NEXT:    fsrm a1
 ; LMULMAX2-RV64F-NEXT:    vsrl.vi v10, v12, 23
 ; LMULMAX2-RV64F-NEXT:    li a1, 127
@@ -1231,24 +1213,22 @@ define void @cttz_v4i64(ptr %x, ptr %y) nounwind {
 ; LMULMAX2-RV32D-LABEL: cttz_v4i64:
 ; LMULMAX2-RV32D:       # %bb.0:
 ; LMULMAX2-RV32D-NEXT:    vsetivli zero, 4, e64, m2, ta, ma
-; LMULMAX2-RV32D-NEXT:    vle64.v v10, (a0)
+; LMULMAX2-RV32D-NEXT:    vle64.v v8, (a0)
 ; LMULMAX2-RV32D-NEXT:    vsetivli zero, 8, e32, m2, ta, ma
-; LMULMAX2-RV32D-NEXT:    vmv.v.i v12, 0
+; LMULMAX2-RV32D-NEXT:    vmv.v.i v10, 0
 ; LMULMAX2-RV32D-NEXT:    vsetivli zero, 4, e64, m2, ta, ma
-; LMULMAX2-RV32D-NEXT:    vmseq.vv v8, v10, v12
-; LMULMAX2-RV32D-NEXT:    vsub.vv v12, v12, v10
-; LMULMAX2-RV32D-NEXT:    vand.vv v10, v10, v12
-; LMULMAX2-RV32D-NEXT:    vmset.m v0
+; LMULMAX2-RV32D-NEXT:    vmseq.vv v0, v8, v10
+; LMULMAX2-RV32D-NEXT:    vsub.vv v10, v10, v8
+; LMULMAX2-RV32D-NEXT:    vand.vv v8, v8, v10
 ; LMULMAX2-RV32D-NEXT:    fsrmi a1, 1
-; LMULMAX2-RV32D-NEXT:    vfcvt.f.xu.v v10, v10, v0.t
+; LMULMAX2-RV32D-NEXT:    vfcvt.f.xu.v v8, v8
 ; LMULMAX2-RV32D-NEXT:    fsrm a1
 ; LMULMAX2-RV32D-NEXT:    li a1, 52
-; LMULMAX2-RV32D-NEXT:    vsrl.vx v10, v10, a1
+; LMULMAX2-RV32D-NEXT:    vsrl.vx v8, v8, a1
 ; LMULMAX2-RV32D-NEXT:    li a1, 1023
-; LMULMAX2-RV32D-NEXT:    vsub.vx v10, v10, a1
+; LMULMAX2-RV32D-NEXT:    vsub.vx v8, v8, a1
 ; LMULMAX2-RV32D-NEXT:    li a1, 64
-; LMULMAX2-RV32D-NEXT:    vmv1r.v v0, v8
-; LMULMAX2-RV32D-NEXT:    vmerge.vxm v8, v10, a1, v0
+; LMULMAX2-RV32D-NEXT:    vmerge.vxm v8, v8, a1, v0
 ; LMULMAX2-RV32D-NEXT:    vse64.v v8, (a0)
 ; LMULMAX2-RV32D-NEXT:    ret
 ;
@@ -1258,9 +1238,8 @@ define void @cttz_v4i64(ptr %x, ptr %y) nounwind {
 ; LMULMAX2-RV64D-NEXT:    vle64.v v8, (a0)
 ; LMULMAX2-RV64D-NEXT:    vrsub.vi v10, v8, 0
 ; LMULMAX2-RV64D-NEXT:    vand.vv v10, v8, v10
-; LMULMAX2-RV64D-NEXT:    vmset.m v0
 ; LMULMAX2-RV64D-NEXT:    fsrmi a1, 1
-; LMULMAX2-RV64D-NEXT:    vfcvt.f.xu.v v10, v10, v0.t
+; LMULMAX2-RV64D-NEXT:    vfcvt.f.xu.v v10, v10
 ; LMULMAX2-RV64D-NEXT:    fsrm a1
 ; LMULMAX2-RV64D-NEXT:    li a1, 52
 ; LMULMAX2-RV64D-NEXT:    vsrl.vx v10, v10, a1
@@ -1275,24 +1254,22 @@ define void @cttz_v4i64(ptr %x, ptr %y) nounwind {
 ; LMULMAX8-RV32-LABEL: cttz_v4i64:
 ; LMULMAX8-RV32:       # %bb.0:
 ; LMULMAX8-RV32-NEXT:    vsetivli zero, 4, e64, m2, ta, ma
-; LMULMAX8-RV32-NEXT:    vle64.v v10, (a0)
+; LMULMAX8-RV32-NEXT:    vle64.v v8, (a0)
 ; LMULMAX8-RV32-NEXT:    vsetivli zero, 8, e32, m2, ta, ma
-; LMULMAX8-RV32-NEXT:    vmv.v.i v12, 0
+; LMULMAX8-RV32-NEXT:    vmv.v.i v10, 0
 ; LMULMAX8-RV32-NEXT:    vsetivli zero, 4, e64, m2, ta, ma
-; LMULMAX8-RV32-NEXT:    vmseq.vv v8, v10, v12
-; LMULMAX8-RV32-NEXT:    vsub.vv v12, v12, v10
-; LMULMAX8-RV32-NEXT:    vand.vv v10, v10, v12
-; LMULMAX8-RV32-NEXT:    vmset.m v0
+; LMULMAX8-RV32-NEXT:    vmseq.vv v0, v8, v10
+; LMULMAX8-RV32-NEXT:    vsub.vv v10, v10, v8
+; LMULMAX8-RV32-NEXT:    vand.vv v8, v8, v10
 ; LMULMAX8-RV32-NEXT:    fsrmi a1, 1
-; LMULMAX8-RV32-NEXT:    vfcvt.f.xu.v v10, v10, v0.t
+; LMULMAX8-RV32-NEXT:    vfcvt.f.xu.v v8, v8
 ; LMULMAX8-RV32-NEXT:    fsrm a1
 ; LMULMAX8-RV32-NEXT:    li a1, 52
-; LMULMAX8-RV32-NEXT:    vsrl.vx v10, v10, a1
+; LMULMAX8-RV32-NEXT:    vsrl.vx v8, v8, a1
 ; LMULMAX8-RV32-NEXT:    li a1, 1023
-; LMULMAX8-RV32-NEXT:    vsub.vx v10, v10, a1
+; LMULMAX8-RV32-NEXT:    vsub.vx v8, v8, a1
 ; LMULMAX8-RV32-NEXT:    li a1, 64
-; LMULMAX8-RV32-NEXT:    vmv1r.v v0, v8
-; LMULMAX8-RV32-NEXT:    vmerge.vxm v8, v10, a1, v0
+; LMULMAX8-RV32-NEXT:    vmerge.vxm v8, v8, a1, v0
 ; LMULMAX8-RV32-NEXT:    vse64.v v8, (a0)
 ; LMULMAX8-RV32-NEXT:    ret
 ;
@@ -1302,9 +1279,8 @@ define void @cttz_v4i64(ptr %x, ptr %y) nounwind {
 ; LMULMAX8-RV64-NEXT:    vle64.v v8, (a0)
 ; LMULMAX8-RV64-NEXT:    vrsub.vi v10, v8, 0
 ; LMULMAX8-RV64-NEXT:    vand.vv v10, v8, v10
-; LMULMAX8-RV64-NEXT:    vmset.m v0
 ; LMULMAX8-RV64-NEXT:    fsrmi a1, 1
-; LMULMAX8-RV64-NEXT:    vfcvt.f.xu.v v10, v10, v0.t
+; LMULMAX8-RV64-NEXT:    vfcvt.f.xu.v v10, v10
 ; LMULMAX8-RV64-NEXT:    fsrm a1
 ; LMULMAX8-RV64-NEXT:    li a1, 52
 ; LMULMAX8-RV64-NEXT:    vsrl.vx v10, v10, a1

diff --git a/llvm/test/CodeGen/RISCV/rvv/float-round-conv.ll b/llvm/test/CodeGen/RISCV/rvv/float-round-conv.ll
index dad9a4f..bad4e74 100644 (file)
--- a/llvm/test/CodeGen/RISCV/rvv/float-round-conv.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/float-round-conv.ll
@@ -487,20 +487,18 @@ define <vscale x 1 x i8> @ceil_nxv1f32_to_ui8(<vscale x 1 x float> %x) {
 define <vscale x 1 x i16> @ceil_nxv1f32_to_si16(<vscale x 1 x float> %x) {
 ; RV32-LABEL: ceil_nxv1f32_to_si16:
 ; RV32:       # %bb.0:
-; RV32-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
-; RV32-NEXT:    vmset.m v0
 ; RV32-NEXT:    fsrmi a0, 3
-; RV32-NEXT:    vfncvt.x.f.w v9, v8, v0.t
+; RV32-NEXT:    vsetvli a1, zero, e16, mf4, ta, ma
+; RV32-NEXT:    vfncvt.x.f.w v9, v8
 ; RV32-NEXT:    fsrm a0
 ; RV32-NEXT:    vmv1r.v v8, v9
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: ceil_nxv1f32_to_si16:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
-; RV64-NEXT:    vmset.m v0
 ; RV64-NEXT:    fsrmi a0, 3
-; RV64-NEXT:    vfncvt.x.f.w v9, v8, v0.t
+; RV64-NEXT:    vsetvli a1, zero, e16, mf4, ta, ma
+; RV64-NEXT:    vfncvt.x.f.w v9, v8
 ; RV64-NEXT:    fsrm a0
 ; RV64-NEXT:    vmv1r.v v8, v9
 ; RV64-NEXT:    ret
@@ -512,20 +510,18 @@ define <vscale x 1 x i16> @ceil_nxv1f32_to_si16(<vscale x 1 x float> %x) {
 define <vscale x 1 x i16> @ceil_nxv1f32_to_ui16(<vscale x 1 x float> %x) {
 ; RV32-LABEL: ceil_nxv1f32_to_ui16:
 ; RV32:       # %bb.0:
-; RV32-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
-; RV32-NEXT:    vmset.m v0
 ; RV32-NEXT:    fsrmi a0, 3
-; RV32-NEXT:    vfncvt.xu.f.w v9, v8, v0.t
+; RV32-NEXT:    vsetvli a1, zero, e16, mf4, ta, ma
+; RV32-NEXT:    vfncvt.xu.f.w v9, v8
 ; RV32-NEXT:    fsrm a0
 ; RV32-NEXT:    vmv1r.v v8, v9
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: ceil_nxv1f32_to_ui16:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
-; RV64-NEXT:    vmset.m v0
 ; RV64-NEXT:    fsrmi a0, 3
-; RV64-NEXT:    vfncvt.xu.f.w v9, v8, v0.t
+; RV64-NEXT:    vsetvli a1, zero, e16, mf4, ta, ma
+; RV64-NEXT:    vfncvt.xu.f.w v9, v8
 ; RV64-NEXT:    fsrm a0
 ; RV64-NEXT:    vmv1r.v v8, v9
 ; RV64-NEXT:    ret
@@ -537,19 +533,17 @@ define <vscale x 1 x i16> @ceil_nxv1f32_to_ui16(<vscale x 1 x float> %x) {
 define <vscale x 1 x i32> @ceil_nxv1f32_to_si32(<vscale x 1 x float> %x) {
 ; RV32-LABEL: ceil_nxv1f32_to_si32:
 ; RV32:       # %bb.0:
-; RV32-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
-; RV32-NEXT:    vmset.m v0
 ; RV32-NEXT:    fsrmi a0, 3
-; RV32-NEXT:    vfcvt.x.f.v v8, v8, v0.t
+; RV32-NEXT:    vsetvli a1, zero, e32, mf2, ta, ma
+; RV32-NEXT:    vfcvt.x.f.v v8, v8
 ; RV32-NEXT:    fsrm a0
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: ceil_nxv1f32_to_si32:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
-; RV64-NEXT:    vmset.m v0
 ; RV64-NEXT:    fsrmi a0, 3
-; RV64-NEXT:    vfcvt.x.f.v v8, v8, v0.t
+; RV64-NEXT:    vsetvli a1, zero, e32, mf2, ta, ma
+; RV64-NEXT:    vfcvt.x.f.v v8, v8
 ; RV64-NEXT:    fsrm a0
 ; RV64-NEXT:    ret
   %a = call <vscale x 1 x float> @llvm.ceil.nxv1f32(<vscale x 1 x float> %x)
@@ -560,19 +554,17 @@ define <vscale x 1 x i32> @ceil_nxv1f32_to_si32(<vscale x 1 x float> %x) {
 define <vscale x 1 x i32> @ceil_nxv1f32_to_ui32(<vscale x 1 x float> %x) {
 ; RV32-LABEL: ceil_nxv1f32_to_ui32:
 ; RV32:       # %bb.0:
-; RV32-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
-; RV32-NEXT:    vmset.m v0
 ; RV32-NEXT:    fsrmi a0, 3
-; RV32-NEXT:    vfcvt.xu.f.v v8, v8, v0.t
+; RV32-NEXT:    vsetvli a1, zero, e32, mf2, ta, ma
+; RV32-NEXT:    vfcvt.xu.f.v v8, v8
 ; RV32-NEXT:    fsrm a0
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: ceil_nxv1f32_to_ui32:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
-; RV64-NEXT:    vmset.m v0
 ; RV64-NEXT:    fsrmi a0, 3
-; RV64-NEXT:    vfcvt.xu.f.v v8, v8, v0.t
+; RV64-NEXT:    vsetvli a1, zero, e32, mf2, ta, ma
+; RV64-NEXT:    vfcvt.xu.f.v v8, v8
 ; RV64-NEXT:    fsrm a0
 ; RV64-NEXT:    ret
   %a = call <vscale x 1 x float> @llvm.ceil.nxv1f32(<vscale x 1 x float> %x)
@@ -583,20 +575,18 @@ define <vscale x 1 x i32> @ceil_nxv1f32_to_ui32(<vscale x 1 x float> %x) {
 define <vscale x 1 x i64> @ceil_nxv1f32_to_si64(<vscale x 1 x float> %x) {
 ; RV32-LABEL: ceil_nxv1f32_to_si64:
 ; RV32:       # %bb.0:
-; RV32-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
-; RV32-NEXT:    vmset.m v0
 ; RV32-NEXT:    fsrmi a0, 3
-; RV32-NEXT:    vfwcvt.x.f.v v9, v8, v0.t
+; RV32-NEXT:    vsetvli a1, zero, e32, mf2, ta, ma
+; RV32-NEXT:    vfwcvt.x.f.v v9, v8
 ; RV32-NEXT:    fsrm a0
 ; RV32-NEXT:    vmv1r.v v8, v9
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: ceil_nxv1f32_to_si64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
-; RV64-NEXT:    vmset.m v0
 ; RV64-NEXT:    fsrmi a0, 3
-; RV64-NEXT:    vfwcvt.x.f.v v9, v8, v0.t
+; RV64-NEXT:    vsetvli a1, zero, e32, mf2, ta, ma
+; RV64-NEXT:    vfwcvt.x.f.v v9, v8
 ; RV64-NEXT:    fsrm a0
 ; RV64-NEXT:    vmv1r.v v8, v9
 ; RV64-NEXT:    ret
@@ -608,20 +598,18 @@ define <vscale x 1 x i64> @ceil_nxv1f32_to_si64(<vscale x 1 x float> %x) {
 define <vscale x 1 x i64> @ceil_nxv1f32_to_ui64(<vscale x 1 x float> %x) {
 ; RV32-LABEL: ceil_nxv1f32_to_ui64:
 ; RV32:       # %bb.0:
-; RV32-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
-; RV32-NEXT:    vmset.m v0
 ; RV32-NEXT:    fsrmi a0, 3
-; RV32-NEXT:    vfwcvt.xu.f.v v9, v8, v0.t
+; RV32-NEXT:    vsetvli a1, zero, e32, mf2, ta, ma
+; RV32-NEXT:    vfwcvt.xu.f.v v9, v8
 ; RV32-NEXT:    fsrm a0
 ; RV32-NEXT:    vmv1r.v v8, v9
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: ceil_nxv1f32_to_ui64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
-; RV64-NEXT:    vmset.m v0
 ; RV64-NEXT:    fsrmi a0, 3
-; RV64-NEXT:    vfwcvt.xu.f.v v9, v8, v0.t
+; RV64-NEXT:    vsetvli a1, zero, e32, mf2, ta, ma
+; RV64-NEXT:    vfwcvt.xu.f.v v9, v8
 ; RV64-NEXT:    fsrm a0
 ; RV64-NEXT:    vmv1r.v v8, v9
 ; RV64-NEXT:    ret
@@ -725,20 +713,18 @@ define <vscale x 4 x i8> @ceil_nxv4f32_to_ui8(<vscale x 4 x float> %x) {
 define <vscale x 4 x i16> @ceil_nxv4f32_to_si16(<vscale x 4 x float> %x) {
 ; RV32-LABEL: ceil_nxv4f32_to_si16:
 ; RV32:       # %bb.0:
-; RV32-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
-; RV32-NEXT:    vmset.m v0
 ; RV32-NEXT:    fsrmi a0, 3
-; RV32-NEXT:    vfncvt.x.f.w v10, v8, v0.t
+; RV32-NEXT:    vsetvli a1, zero, e16, m1, ta, ma
+; RV32-NEXT:    vfncvt.x.f.w v10, v8
 ; RV32-NEXT:    fsrm a0
 ; RV32-NEXT:    vmv.v.v v8, v10
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: ceil_nxv4f32_to_si16:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
-; RV64-NEXT:    vmset.m v0
 ; RV64-NEXT:    fsrmi a0, 3
-; RV64-NEXT:    vfncvt.x.f.w v10, v8, v0.t
+; RV64-NEXT:    vsetvli a1, zero, e16, m1, ta, ma
+; RV64-NEXT:    vfncvt.x.f.w v10, v8
 ; RV64-NEXT:    fsrm a0
 ; RV64-NEXT:    vmv.v.v v8, v10
 ; RV64-NEXT:    ret
@@ -750,20 +736,18 @@ define <vscale x 4 x i16> @ceil_nxv4f32_to_si16(<vscale x 4 x float> %x) {
 define <vscale x 4 x i16> @ceil_nxv4f32_to_ui16(<vscale x 4 x float> %x) {
 ; RV32-LABEL: ceil_nxv4f32_to_ui16:
 ; RV32:       # %bb.0:
-; RV32-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
-; RV32-NEXT:    vmset.m v0
 ; RV32-NEXT:    fsrmi a0, 3
-; RV32-NEXT:    vfncvt.xu.f.w v10, v8, v0.t
+; RV32-NEXT:    vsetvli a1, zero, e16, m1, ta, ma
+; RV32-NEXT:    vfncvt.xu.f.w v10, v8
 ; RV32-NEXT:    fsrm a0
 ; RV32-NEXT:    vmv.v.v v8, v10
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: ceil_nxv4f32_to_ui16:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
-; RV64-NEXT:    vmset.m v0
 ; RV64-NEXT:    fsrmi a0, 3
-; RV64-NEXT:    vfncvt.xu.f.w v10, v8, v0.t
+; RV64-NEXT:    vsetvli a1, zero, e16, m1, ta, ma
+; RV64-NEXT:    vfncvt.xu.f.w v10, v8
 ; RV64-NEXT:    fsrm a0
 ; RV64-NEXT:    vmv.v.v v8, v10
 ; RV64-NEXT:    ret
@@ -775,19 +759,17 @@ define <vscale x 4 x i16> @ceil_nxv4f32_to_ui16(<vscale x 4 x float> %x) {
 define <vscale x 4 x i32> @ceil_nxv4f32_to_si32(<vscale x 4 x float> %x) {
 ; RV32-LABEL: ceil_nxv4f32_to_si32:
 ; RV32:       # %bb.0:
-; RV32-NEXT:    vsetvli a0, zero, e32, m2, ta, ma
-; RV32-NEXT:    vmset.m v0
 ; RV32-NEXT:    fsrmi a0, 3
-; RV32-NEXT:    vfcvt.x.f.v v8, v8, v0.t
+; RV32-NEXT:    vsetvli a1, zero, e32, m2, ta, ma
+; RV32-NEXT:    vfcvt.x.f.v v8, v8
 ; RV32-NEXT:    fsrm a0
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: ceil_nxv4f32_to_si32:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a0, zero, e32, m2, ta, ma
-; RV64-NEXT:    vmset.m v0
 ; RV64-NEXT:    fsrmi a0, 3
-; RV64-NEXT:    vfcvt.x.f.v v8, v8, v0.t
+; RV64-NEXT:    vsetvli a1, zero, e32, m2, ta, ma
+; RV64-NEXT:    vfcvt.x.f.v v8, v8
 ; RV64-NEXT:    fsrm a0
 ; RV64-NEXT:    ret
   %a = call <vscale x 4 x float> @llvm.ceil.nxv4f32(<vscale x 4 x float> %x)
@@ -798,19 +780,17 @@ define <vscale x 4 x i32> @ceil_nxv4f32_to_si32(<vscale x 4 x float> %x) {
 define <vscale x 4 x i32> @ceil_nxv4f32_to_ui32(<vscale x 4 x float> %x) {
 ; RV32-LABEL: ceil_nxv4f32_to_ui32:
 ; RV32:       # %bb.0:
-; RV32-NEXT:    vsetvli a0, zero, e32, m2, ta, ma
-; RV32-NEXT:    vmset.m v0
 ; RV32-NEXT:    fsrmi a0, 3
-; RV32-NEXT:    vfcvt.xu.f.v v8, v8, v0.t
+; RV32-NEXT:    vsetvli a1, zero, e32, m2, ta, ma
+; RV32-NEXT:    vfcvt.xu.f.v v8, v8
 ; RV32-NEXT:    fsrm a0
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: ceil_nxv4f32_to_ui32:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a0, zero, e32, m2, ta, ma
-; RV64-NEXT:    vmset.m v0
 ; RV64-NEXT:    fsrmi a0, 3
-; RV64-NEXT:    vfcvt.xu.f.v v8, v8, v0.t
+; RV64-NEXT:    vsetvli a1, zero, e32, m2, ta, ma
+; RV64-NEXT:    vfcvt.xu.f.v v8, v8
 ; RV64-NEXT:    fsrm a0
 ; RV64-NEXT:    ret
   %a = call <vscale x 4 x float> @llvm.ceil.nxv4f32(<vscale x 4 x float> %x)
@@ -821,20 +801,18 @@ define <vscale x 4 x i32> @ceil_nxv4f32_to_ui32(<vscale x 4 x float> %x) {
 define <vscale x 4 x i64> @ceil_nxv4f32_to_si64(<vscale x 4 x float> %x) {
 ; RV32-LABEL: ceil_nxv4f32_to_si64:
 ; RV32:       # %bb.0:
-; RV32-NEXT:    vsetvli a0, zero, e32, m2, ta, ma
-; RV32-NEXT:    vmset.m v0
 ; RV32-NEXT:    fsrmi a0, 3
-; RV32-NEXT:    vfwcvt.x.f.v v12, v8, v0.t
+; RV32-NEXT:    vsetvli a1, zero, e32, m2, ta, ma
+; RV32-NEXT:    vfwcvt.x.f.v v12, v8
 ; RV32-NEXT:    fsrm a0
 ; RV32-NEXT:    vmv4r.v v8, v12
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: ceil_nxv4f32_to_si64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a0, zero, e32, m2, ta, ma
-; RV64-NEXT:    vmset.m v0
 ; RV64-NEXT:    fsrmi a0, 3
-; RV64-NEXT:    vfwcvt.x.f.v v12, v8, v0.t
+; RV64-NEXT:    vsetvli a1, zero, e32, m2, ta, ma
+; RV64-NEXT:    vfwcvt.x.f.v v12, v8
 ; RV64-NEXT:    fsrm a0
 ; RV64-NEXT:    vmv4r.v v8, v12
 ; RV64-NEXT:    ret
@@ -846,20 +824,18 @@ define <vscale x 4 x i64> @ceil_nxv4f32_to_si64(<vscale x 4 x float> %x) {
 define <vscale x 4 x i64> @ceil_nxv4f32_to_ui64(<vscale x 4 x float> %x) {
 ; RV32-LABEL: ceil_nxv4f32_to_ui64:
 ; RV32:       # %bb.0:
-; RV32-NEXT:    vsetvli a0, zero, e32, m2, ta, ma
-; RV32-NEXT:    vmset.m v0
 ; RV32-NEXT:    fsrmi a0, 3
-; RV32-NEXT:    vfwcvt.xu.f.v v12, v8, v0.t
+; RV32-NEXT:    vsetvli a1, zero, e32, m2, ta, ma
+; RV32-NEXT:    vfwcvt.xu.f.v v12, v8
 ; RV32-NEXT:    fsrm a0
 ; RV32-NEXT:    vmv4r.v v8, v12
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: ceil_nxv4f32_to_ui64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a0, zero, e32, m2, ta, ma
-; RV64-NEXT:    vmset.m v0
 ; RV64-NEXT:    fsrmi a0, 3
-; RV64-NEXT:    vfwcvt.xu.f.v v12, v8, v0.t
+; RV64-NEXT:    vsetvli a1, zero, e32, m2, ta, ma
+; RV64-NEXT:    vfwcvt.xu.f.v v12, v8
 ; RV64-NEXT:    fsrm a0
 ; RV64-NEXT:    vmv4r.v v8, v12
 ; RV64-NEXT:    ret

diff --git a/llvm/test/CodeGen/RISCV/rvv/half-round-conv.ll b/llvm/test/CodeGen/RISCV/rvv/half-round-conv.ll
index 4ba88a9..c8ff6b4 100644 (file)
--- a/llvm/test/CodeGen/RISCV/rvv/half-round-conv.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/half-round-conv.ll
@@ -393,20 +393,18 @@ declare <vscale x 1 x half> @llvm.ceil.nxv1f16(<vscale x 1 x half>)
 define <vscale x 1 x i8> @ceil_nxv1f16_to_si8(<vscale x 1 x half> %x) {
 ; RV32-LABEL: ceil_nxv1f16_to_si8:
 ; RV32:       # %bb.0:
-; RV32-NEXT:    vsetvli a0, zero, e8, mf8, ta, ma
-; RV32-NEXT:    vmset.m v0
 ; RV32-NEXT:    fsrmi a0, 3
-; RV32-NEXT:    vfncvt.x.f.w v9, v8, v0.t
+; RV32-NEXT:    vsetvli a1, zero, e8, mf8, ta, ma
+; RV32-NEXT:    vfncvt.x.f.w v9, v8
 ; RV32-NEXT:    fsrm a0
 ; RV32-NEXT:    vmv1r.v v8, v9
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: ceil_nxv1f16_to_si8:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a0, zero, e8, mf8, ta, ma
-; RV64-NEXT:    vmset.m v0
 ; RV64-NEXT:    fsrmi a0, 3
-; RV64-NEXT:    vfncvt.x.f.w v9, v8, v0.t
+; RV64-NEXT:    vsetvli a1, zero, e8, mf8, ta, ma
+; RV64-NEXT:    vfncvt.x.f.w v9, v8
 ; RV64-NEXT:    fsrm a0
 ; RV64-NEXT:    vmv1r.v v8, v9
 ; RV64-NEXT:    ret
@@ -418,20 +416,18 @@ define <vscale x 1 x i8> @ceil_nxv1f16_to_si8(<vscale x 1 x half> %x) {
 define <vscale x 1 x i8> @ceil_nxv1f16_to_ui8(<vscale x 1 x half> %x) {
 ; RV32-LABEL: ceil_nxv1f16_to_ui8:
 ; RV32:       # %bb.0:
-; RV32-NEXT:    vsetvli a0, zero, e8, mf8, ta, ma
-; RV32-NEXT:    vmset.m v0
 ; RV32-NEXT:    fsrmi a0, 3
-; RV32-NEXT:    vfncvt.xu.f.w v9, v8, v0.t
+; RV32-NEXT:    vsetvli a1, zero, e8, mf8, ta, ma
+; RV32-NEXT:    vfncvt.xu.f.w v9, v8
 ; RV32-NEXT:    fsrm a0
 ; RV32-NEXT:    vmv1r.v v8, v9
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: ceil_nxv1f16_to_ui8:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a0, zero, e8, mf8, ta, ma
-; RV64-NEXT:    vmset.m v0
 ; RV64-NEXT:    fsrmi a0, 3
-; RV64-NEXT:    vfncvt.xu.f.w v9, v8, v0.t
+; RV64-NEXT:    vsetvli a1, zero, e8, mf8, ta, ma
+; RV64-NEXT:    vfncvt.xu.f.w v9, v8
 ; RV64-NEXT:    fsrm a0
 ; RV64-NEXT:    vmv1r.v v8, v9
 ; RV64-NEXT:    ret
@@ -443,19 +439,17 @@ define <vscale x 1 x i8> @ceil_nxv1f16_to_ui8(<vscale x 1 x half> %x) {
 define <vscale x 1 x i16> @ceil_nxv1f16_to_si16(<vscale x 1 x half> %x) {
 ; RV32-LABEL: ceil_nxv1f16_to_si16:
 ; RV32:       # %bb.0:
-; RV32-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
-; RV32-NEXT:    vmset.m v0
 ; RV32-NEXT:    fsrmi a0, 3
-; RV32-NEXT:    vfcvt.x.f.v v8, v8, v0.t
+; RV32-NEXT:    vsetvli a1, zero, e16, mf4, ta, ma
+; RV32-NEXT:    vfcvt.x.f.v v8, v8
 ; RV32-NEXT:    fsrm a0
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: ceil_nxv1f16_to_si16:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
-; RV64-NEXT:    vmset.m v0
 ; RV64-NEXT:    fsrmi a0, 3
-; RV64-NEXT:    vfcvt.x.f.v v8, v8, v0.t
+; RV64-NEXT:    vsetvli a1, zero, e16, mf4, ta, ma
+; RV64-NEXT:    vfcvt.x.f.v v8, v8
 ; RV64-NEXT:    fsrm a0
 ; RV64-NEXT:    ret
   %a = call <vscale x 1 x half> @llvm.ceil.nxv1f16(<vscale x 1 x half> %x)
@@ -466,19 +460,17 @@ define <vscale x 1 x i16> @ceil_nxv1f16_to_si16(<vscale x 1 x half> %x) {
 define <vscale x 1 x i16> @ceil_nxv1f16_to_ui16(<vscale x 1 x half> %x) {
 ; RV32-LABEL: ceil_nxv1f16_to_ui16:
 ; RV32:       # %bb.0:
-; RV32-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
-; RV32-NEXT:    vmset.m v0
 ; RV32-NEXT:    fsrmi a0, 3
-; RV32-NEXT:    vfcvt.xu.f.v v8, v8, v0.t
+; RV32-NEXT:    vsetvli a1, zero, e16, mf4, ta, ma
+; RV32-NEXT:    vfcvt.xu.f.v v8, v8
 ; RV32-NEXT:    fsrm a0
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: ceil_nxv1f16_to_ui16:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
-; RV64-NEXT:    vmset.m v0
 ; RV64-NEXT:    fsrmi a0, 3
-; RV64-NEXT:    vfcvt.xu.f.v v8, v8, v0.t
+; RV64-NEXT:    vsetvli a1, zero, e16, mf4, ta, ma
+; RV64-NEXT:    vfcvt.xu.f.v v8, v8
 ; RV64-NEXT:    fsrm a0
 ; RV64-NEXT:    ret
   %a = call <vscale x 1 x half> @llvm.ceil.nxv1f16(<vscale x 1 x half> %x)
@@ -489,20 +481,18 @@ define <vscale x 1 x i16> @ceil_nxv1f16_to_ui16(<vscale x 1 x half> %x) {
 define <vscale x 1 x i32> @ceil_nxv1f16_to_si32(<vscale x 1 x half> %x) {
 ; RV32-LABEL: ceil_nxv1f16_to_si32:
 ; RV32:       # %bb.0:
-; RV32-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
-; RV32-NEXT:    vmset.m v0
 ; RV32-NEXT:    fsrmi a0, 3
-; RV32-NEXT:    vfwcvt.x.f.v v9, v8, v0.t
+; RV32-NEXT:    vsetvli a1, zero, e16, mf4, ta, ma
+; RV32-NEXT:    vfwcvt.x.f.v v9, v8
 ; RV32-NEXT:    fsrm a0
 ; RV32-NEXT:    vmv1r.v v8, v9
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: ceil_nxv1f16_to_si32:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
-; RV64-NEXT:    vmset.m v0
 ; RV64-NEXT:    fsrmi a0, 3
-; RV64-NEXT:    vfwcvt.x.f.v v9, v8, v0.t
+; RV64-NEXT:    vsetvli a1, zero, e16, mf4, ta, ma
+; RV64-NEXT:    vfwcvt.x.f.v v9, v8
 ; RV64-NEXT:    fsrm a0
 ; RV64-NEXT:    vmv1r.v v8, v9
 ; RV64-NEXT:    ret
@@ -514,20 +504,18 @@ define <vscale x 1 x i32> @ceil_nxv1f16_to_si32(<vscale x 1 x half> %x) {
 define <vscale x 1 x i32> @ceil_nxv1f16_to_ui32(<vscale x 1 x half> %x) {
 ; RV32-LABEL: ceil_nxv1f16_to_ui32:
 ; RV32:       # %bb.0:
-; RV32-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
-; RV32-NEXT:    vmset.m v0
 ; RV32-NEXT:    fsrmi a0, 3
-; RV32-NEXT:    vfwcvt.xu.f.v v9, v8, v0.t
+; RV32-NEXT:    vsetvli a1, zero, e16, mf4, ta, ma
+; RV32-NEXT:    vfwcvt.xu.f.v v9, v8
 ; RV32-NEXT:    fsrm a0
 ; RV32-NEXT:    vmv1r.v v8, v9
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: ceil_nxv1f16_to_ui32:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
-; RV64-NEXT:    vmset.m v0
 ; RV64-NEXT:    fsrmi a0, 3
-; RV64-NEXT:    vfwcvt.xu.f.v v9, v8, v0.t
+; RV64-NEXT:    vsetvli a1, zero, e16, mf4, ta, ma
+; RV64-NEXT:    vfwcvt.xu.f.v v9, v8
 ; RV64-NEXT:    fsrm a0
 ; RV64-NEXT:    vmv1r.v v8, v9
 ; RV64-NEXT:    ret
@@ -627,20 +615,18 @@ declare <vscale x 4 x half> @llvm.ceil.nxv4f16(<vscale x 4 x half>)
 define <vscale x 4 x i8> @ceil_nxv4f16_to_si8(<vscale x 4 x half> %x) {
 ; RV32-LABEL: ceil_nxv4f16_to_si8:
 ; RV32:       # %bb.0:
-; RV32-NEXT:    vsetvli a0, zero, e8, mf2, ta, ma
-; RV32-NEXT:    vmset.m v0
 ; RV32-NEXT:    fsrmi a0, 3
-; RV32-NEXT:    vfncvt.x.f.w v9, v8, v0.t
+; RV32-NEXT:    vsetvli a1, zero, e8, mf2, ta, ma
+; RV32-NEXT:    vfncvt.x.f.w v9, v8
 ; RV32-NEXT:    fsrm a0
 ; RV32-NEXT:    vmv1r.v v8, v9
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: ceil_nxv4f16_to_si8:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a0, zero, e8, mf2, ta, ma
-; RV64-NEXT:    vmset.m v0
 ; RV64-NEXT:    fsrmi a0, 3
-; RV64-NEXT:    vfncvt.x.f.w v9, v8, v0.t
+; RV64-NEXT:    vsetvli a1, zero, e8, mf2, ta, ma
+; RV64-NEXT:    vfncvt.x.f.w v9, v8
 ; RV64-NEXT:    fsrm a0
 ; RV64-NEXT:    vmv1r.v v8, v9
 ; RV64-NEXT:    ret
@@ -652,20 +638,18 @@ define <vscale x 4 x i8> @ceil_nxv4f16_to_si8(<vscale x 4 x half> %x) {
 define <vscale x 4 x i8> @ceil_nxv4f16_to_ui8(<vscale x 4 x half> %x) {
 ; RV32-LABEL: ceil_nxv4f16_to_ui8:
 ; RV32:       # %bb.0:
-; RV32-NEXT:    vsetvli a0, zero, e8, mf2, ta, ma
-; RV32-NEXT:    vmset.m v0
 ; RV32-NEXT:    fsrmi a0, 3
-; RV32-NEXT:    vfncvt.xu.f.w v9, v8, v0.t
+; RV32-NEXT:    vsetvli a1, zero, e8, mf2, ta, ma
+; RV32-NEXT:    vfncvt.xu.f.w v9, v8
 ; RV32-NEXT:    fsrm a0
 ; RV32-NEXT:    vmv1r.v v8, v9
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: ceil_nxv4f16_to_ui8:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a0, zero, e8, mf2, ta, ma
-; RV64-NEXT:    vmset.m v0
 ; RV64-NEXT:    fsrmi a0, 3
-; RV64-NEXT:    vfncvt.xu.f.w v9, v8, v0.t
+; RV64-NEXT:    vsetvli a1, zero, e8, mf2, ta, ma
+; RV64-NEXT:    vfncvt.xu.f.w v9, v8
 ; RV64-NEXT:    fsrm a0
 ; RV64-NEXT:    vmv1r.v v8, v9
 ; RV64-NEXT:    ret
@@ -677,19 +661,17 @@ define <vscale x 4 x i8> @ceil_nxv4f16_to_ui8(<vscale x 4 x half> %x) {
 define <vscale x 4 x i16> @ceil_nxv4f16_to_si16(<vscale x 4 x half> %x) {
 ; RV32-LABEL: ceil_nxv4f16_to_si16:
 ; RV32:       # %bb.0:
-; RV32-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
-; RV32-NEXT:    vmset.m v0
 ; RV32-NEXT:    fsrmi a0, 3
-; RV32-NEXT:    vfcvt.x.f.v v8, v8, v0.t
+; RV32-NEXT:    vsetvli a1, zero, e16, m1, ta, ma
+; RV32-NEXT:    vfcvt.x.f.v v8, v8
 ; RV32-NEXT:    fsrm a0
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: ceil_nxv4f16_to_si16:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
-; RV64-NEXT:    vmset.m v0
 ; RV64-NEXT:    fsrmi a0, 3
-; RV64-NEXT:    vfcvt.x.f.v v8, v8, v0.t
+; RV64-NEXT:    vsetvli a1, zero, e16, m1, ta, ma
+; RV64-NEXT:    vfcvt.x.f.v v8, v8
 ; RV64-NEXT:    fsrm a0
 ; RV64-NEXT:    ret
   %a = call <vscale x 4 x half> @llvm.ceil.nxv4f16(<vscale x 4 x half> %x)
@@ -700,19 +682,17 @@ define <vscale x 4 x i16> @ceil_nxv4f16_to_si16(<vscale x 4 x half> %x) {
 define <vscale x 4 x i16> @ceil_nxv4f16_to_ui16(<vscale x 4 x half> %x) {
 ; RV32-LABEL: ceil_nxv4f16_to_ui16:
 ; RV32:       # %bb.0:
-; RV32-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
-; RV32-NEXT:    vmset.m v0
 ; RV32-NEXT:    fsrmi a0, 3
-; RV32-NEXT:    vfcvt.xu.f.v v8, v8, v0.t
+; RV32-NEXT:    vsetvli a1, zero, e16, m1, ta, ma
+; RV32-NEXT:    vfcvt.xu.f.v v8, v8
 ; RV32-NEXT:    fsrm a0
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: ceil_nxv4f16_to_ui16:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
-; RV64-NEXT:    vmset.m v0
 ; RV64-NEXT:    fsrmi a0, 3
-; RV64-NEXT:    vfcvt.xu.f.v v8, v8, v0.t
+; RV64-NEXT:    vsetvli a1, zero, e16, m1, ta, ma
+; RV64-NEXT:    vfcvt.xu.f.v v8, v8
 ; RV64-NEXT:    fsrm a0
 ; RV64-NEXT:    ret
   %a = call <vscale x 4 x half> @llvm.ceil.nxv4f16(<vscale x 4 x half> %x)
@@ -723,20 +703,18 @@ define <vscale x 4 x i16> @ceil_nxv4f16_to_ui16(<vscale x 4 x half> %x) {
 define <vscale x 4 x i32> @ceil_nxv4f16_to_si32(<vscale x 4 x half> %x) {
 ; RV32-LABEL: ceil_nxv4f16_to_si32:
 ; RV32:       # %bb.0:
-; RV32-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
-; RV32-NEXT:    vmset.m v0
 ; RV32-NEXT:    fsrmi a0, 3
-; RV32-NEXT:    vfwcvt.x.f.v v10, v8, v0.t
+; RV32-NEXT:    vsetvli a1, zero, e16, m1, ta, ma
+; RV32-NEXT:    vfwcvt.x.f.v v10, v8
 ; RV32-NEXT:    fsrm a0
 ; RV32-NEXT:    vmv2r.v v8, v10
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: ceil_nxv4f16_to_si32:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
-; RV64-NEXT:    vmset.m v0
 ; RV64-NEXT:    fsrmi a0, 3
-; RV64-NEXT:    vfwcvt.x.f.v v10, v8, v0.t
+; RV64-NEXT:    vsetvli a1, zero, e16, m1, ta, ma
+; RV64-NEXT:    vfwcvt.x.f.v v10, v8
 ; RV64-NEXT:    fsrm a0
 ; RV64-NEXT:    vmv2r.v v8, v10
 ; RV64-NEXT:    ret
@@ -748,20 +726,18 @@ define <vscale x 4 x i32> @ceil_nxv4f16_to_si32(<vscale x 4 x half> %x) {
 define <vscale x 4 x i32> @ceil_nxv4f16_to_ui32(<vscale x 4 x half> %x) {
 ; RV32-LABEL: ceil_nxv4f16_to_ui32:
 ; RV32:       # %bb.0:
-; RV32-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
-; RV32-NEXT:    vmset.m v0
 ; RV32-NEXT:    fsrmi a0, 3
-; RV32-NEXT:    vfwcvt.xu.f.v v10, v8, v0.t
+; RV32-NEXT:    vsetvli a1, zero, e16, m1, ta, ma
+; RV32-NEXT:    vfwcvt.xu.f.v v10, v8
 ; RV32-NEXT:    fsrm a0
 ; RV32-NEXT:    vmv2r.v v8, v10
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: ceil_nxv4f16_to_ui32:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
-; RV64-NEXT:    vmset.m v0
 ; RV64-NEXT:    fsrmi a0, 3
-; RV64-NEXT:    vfwcvt.xu.f.v v10, v8, v0.t
+; RV64-NEXT:    vsetvli a1, zero, e16, m1, ta, ma
+; RV64-NEXT:    vfwcvt.xu.f.v v10, v8
 ; RV64-NEXT:    fsrm a0
 ; RV64-NEXT:    vmv2r.v v8, v10
 ; RV64-NEXT:    ret