[RISCV] Handle zvbb instructions in getVectorLowDemandedScalarBits. (#129011)

Author: tclin914

llvm/lib/Target/RISCV/RISCVInstrInfo.cpp

@@ -4107,7 +4107,6 @@ bool RISCV::hasEqualFRM(const MachineInstr &MI1, const MachineInstr &MI2) {
 
 std::optional<unsigned>
 RISCV::getVectorLowDemandedScalarBits(uint16_t Opcode, unsigned Log2SEW) {
-  // TODO: Handle Zvbb instructions
   switch (Opcode) {
   default:
     return std::nullopt;
@@ -4119,6 +4118,9 @@ RISCV::getVectorLowDemandedScalarBits(uint16_t Opcode, unsigned Log2SEW) {
   // 12.4. Vector Single-Width Scaling Shift Instructions
   case RISCV::VSSRL_VX:
   case RISCV::VSSRA_VX:
+  // Zvbb
+  case RISCV::VROL_VX:
+  case RISCV::VROR_VX:
     // Only the low lg2(SEW) bits of the shift-amount value are used.
     return Log2SEW;
 
@@ -4128,6 +4130,8 @@ RISCV::getVectorLowDemandedScalarBits(uint16_t Opcode, unsigned Log2SEW) {
   // 12.5. Vector Narrowing Fixed-Point Clip Instructions
   case RISCV::VNCLIPU_WX:
   case RISCV::VNCLIP_WX:
+  // Zvbb
+  case RISCV::VWSLL_VX:
     // Only the low lg2(2*SEW) bits of the shift-amount value are used.
     return Log2SEW + 1;
 
@@ -4213,6 +4217,8 @@ RISCV::getVectorLowDemandedScalarBits(uint16_t Opcode, unsigned Log2SEW) {
   case RISCV::VSMUL_VX:
   // 16.1. Integer Scalar Move Instructions
   case RISCV::VMV_S_X:
+  // Zvbb
+  case RISCV::VANDN_VX:
     return 1U << Log2SEW;
   }
 }

llvm/test/CodeGen/RISCV/rvv/zvbb-demanded-bits.ll

@@ -0,0 +1,63 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 5
+; RUN: llc -mtriple=riscv64 -mattr=+v,+zvbb,+prefer-w-inst \
+; RUN:   -verify-machineinstrs < %s | FileCheck %s
+
+define <vscale x 1 x i8> @vrol_vx_nxv1i8(<vscale x 1 x i8> %a, i8 %b) {
+; CHECK-LABEL: vrol_vx_nxv1i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addiw a0, a0, 1
+; CHECK-NEXT:    vsetvli a1, zero, e8, mf8, ta, ma
+; CHECK-NEXT:    vrol.vx v8, v8, a0
+; CHECK-NEXT:    ret
+  %s = add i8 %b, 1
+  %b.head = insertelement <vscale x 1 x i8> poison, i8 %s, i32 0
+  %b.splat = shufflevector <vscale x 1 x i8> %b.head, <vscale x 1 x i8> poison, <vscale x 1 x i32> zeroinitializer
+  %x = call <vscale x 1 x i8> @llvm.fshl.nxv1i8(<vscale x 1 x i8> %a, <vscale x 1 x i8> %a, <vscale x 1 x i8> %b.splat)
+  ret <vscale x 1 x i8> %x
+}
+
+define <vscale x 1 x i8> @vror_vx_nxv1i8(<vscale x 1 x i8> %a, i8 %b) {
+; CHECK-LABEL: vror_vx_nxv1i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addiw a0, a0, 1
+; CHECK-NEXT:    vsetvli a1, zero, e8, mf8, ta, ma
+; CHECK-NEXT:    vror.vx v8, v8, a0
+; CHECK-NEXT:    ret
+  %s = add i8 %b, 1
+  %b.head = insertelement <vscale x 1 x i8> poison, i8 %s, i32 0
+  %b.splat = shufflevector <vscale x 1 x i8> %b.head, <vscale x 1 x i8> poison, <vscale x 1 x i32> zeroinitializer
+  %x = call <vscale x 1 x i8> @llvm.fshr.nxv1i8(<vscale x 1 x i8> %a, <vscale x 1 x i8> %a, <vscale x 1 x i8> %b.splat)
+  ret <vscale x 1 x i8> %x
+}
+
+define <vscale x 2 x i64> @vwsll_vx_i8_nxv2i64_zext(<vscale x 2 x i32> %a, i8 %b) {
+; CHECK-LABEL: vwsll_vx_i8_nxv2i64_zext:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addiw a0, a0, 1
+; CHECK-NEXT:    vsetvli a1, zero, e32, m1, ta, ma
+; CHECK-NEXT:    vwsll.vx v10, v8, a0
+; CHECK-NEXT:    vmv2r.v v8, v10
+; CHECK-NEXT:    ret
+  %s = add i8 %b, 1
+  %head = insertelement <vscale x 2 x i8> poison, i8 %s, i32 0
+  %splat = shufflevector <vscale x 2 x i8> %head, <vscale x 2 x i8> poison, <vscale x 2 x i32> zeroinitializer
+  %x = zext <vscale x 2 x i32> %a to <vscale x 2 x i64>
+  %y = zext <vscale x 2 x i8> %splat to <vscale x 2 x i64>
+  %z = shl <vscale x 2 x i64> %x, %y
+  ret <vscale x 2 x i64> %z
+}
+
+define <vscale x 1 x i8> @vandn_vx_nxv1i8(i8 %x, <vscale x 1 x i8> %y) {
+; CHECK-LABEL: vandn_vx_nxv1i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addiw a0, a0, 1
+; CHECK-NEXT:    vsetvli a1, zero, e8, mf8, ta, ma
+; CHECK-NEXT:    vandn.vx v8, v8, a0
+; CHECK-NEXT:    ret
+  %s = add i8 %x, 1
+  %a = xor i8 %s, -1
+  %head = insertelement <vscale x 1 x i8> poison, i8 %a, i32 0
+  %splat = shufflevector <vscale x 1 x i8> %head, <vscale x 1 x i8> poison, <vscale x 1 x i32> zeroinitializer
+  %b = and <vscale x 1 x i8> %splat, %y
+  ret <vscale x 1 x i8> %b
+}