src/hotspot/cpu/ppc/matcher_ppc.hpp

/*
 * Copyright (c) 2021, 2023, Oracle and/or its affiliates. All rights reserved.
 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
 *
 * This code is free software; you can redistribute it and/or modify it
 * under the terms of the GNU General Public License version 2 only, as
 * published by the Free Software Foundation.
 *
 * This code is distributed in the hope that it will be useful, but WITHOUT
 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
 * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
 * version 2 for more details (a copy is included in the LICENSE file that
 * accompanied this code).
 *
 * You should have received a copy of the GNU General Public License version
 * 2 along with this work; if not, write to the Free Software Foundation,
 * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
 *
 * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
 * or visit www.oracle.com if you need additional information or have any
 * questions.
 *
 */

#ifndef CPU_PPC_MATCHER_PPC_HPP
#define CPU_PPC_MATCHER_PPC_HPP

  // Defined within class Matcher

  // false => size gets scaled to BytesPerLong, ok.
  static const bool init_array_count_is_in_bytes = false;

  // Whether this platform implements the scalable vector feature
  static const bool implements_scalable_vector = false;

  static constexpr bool supports_scalable_vector() {
    return false;
  }

  // PPC implementation uses VSX load/store instructions (if
  // SuperwordUseVSX) which support 4 byte but not arbitrary alignment
  static const bool misaligned_vectors_ok() {
    return false;
  }

  // Whether code generation need accurate ConvI2L types.
  static const bool convi2l_type_required = true;

  // Do we need to mask the count passed to shift instructions or does
  // the cpu only look at the lower 5/6 bits anyway?
  // PowerPC requires masked shift counts.
  static const bool need_masked_shift_count = true;

  // Power6 requires postalloc expand (see block.cpp for description of postalloc expand).
  static const bool require_postalloc_expand = true;

  // No support for generic vector operands.
  static const bool supports_generic_vector_operands = false;

  static constexpr bool isSimpleConstant64(jlong value) {
    // Probably always true, even if a temp register is required.
    return true;
  }

  // Use conditional move (CMOVL) on Power7.
  static constexpr int long_cmove_cost() { return 0; } // this only makes long cmoves more expensive than int cmoves

  // Suppress CMOVF. Conditional move available (sort of) on PPC64 only from P7 onwards. Not exploited yet.
  // fsel doesn't accept a condition register as input, so this would be slightly different.
  static int float_cmove_cost() { return ConditionalMoveLimit; }

  // This affects two different things:
  //  - how Decode nodes are matched
  //  - how ImplicitNullCheck opportunities are recognized
  // If true, the matcher will try to remove all Decodes and match them
  // (as operands) into nodes. NullChecks are not prepared to deal with
  // Decodes by final_graph_reshaping().
  // If false, final_graph_reshaping() forces the decode behind the Cmp
  // for a NullCheck. The matcher matches the Decode node into a register.
  // Implicit_null_check optimization moves the Decode along with the
  // memory operation back up before the NullCheck.
  static constexpr bool narrow_oop_use_complex_address() {
    // TODO: PPC port if (MatchDecodeNodes) return true;
    return false;
  }

  static bool narrow_klass_use_complex_address() {
    NOT_LP64(ShouldNotCallThis());
    assert(UseCompressedClassPointers, "only for compressed klass code");
    // TODO: PPC port if (MatchDecodeNodes) return true;
    return false;
  }

  static bool const_oop_prefer_decode() {
    // Prefer ConN+DecodeN over ConP in simple compressed oops mode.
    return CompressedOops::base() == nullptr;
  }

  static bool const_klass_prefer_decode() {
    // Prefer ConNKlass+DecodeNKlass over ConP in simple compressed klass mode.
    return CompressedKlassPointers::base() == nullptr;
  }

  // Is it better to copy float constants, or load them directly from memory?
  // Intel can load a float constant from a direct address, requiring no
  // extra registers. Most RISCs will have to materialize an address into a
  // register first, so they would do better to copy the constant from stack.
  static const bool rematerialize_float_constants = false;

  // If CPU can load and store mis-aligned doubles directly then no fixup is
  // needed. Else we split the double into 2 integer pieces and move it
  // piece-by-piece. Only happens when passing doubles into C code as the
  // Java calling convention forces doubles to be aligned.
  static const bool misaligned_doubles_ok = true;

  // Advertise here if the CPU requires explicit rounding operations to implement strictfp mode.
  static const bool strict_fp_requires_explicit_rounding = false;

  // Do floats take an entire double register or just half?
  //
  // A float occupies a ppc64 double register. For the allocator, a
  // ppc64 double register appears as a pair of float registers.
  static constexpr bool float_in_double() { return true; }

  // Do ints take an entire long register or just half?
  // The relevant question is how the int is callee-saved:
  // the whole long is written but de-opt'ing will have to extract
  // the relevant 32 bits.
  static const bool int_in_long = true;

  // Does the CPU supports vector variable shift instructions?
  static constexpr bool supports_vector_variable_shifts(void) {
    return false;
  }

  // Does the CPU supports vector variable rotate instructions?
  static constexpr bool supports_vector_variable_rotates(void) {
    return false;
  }

  // Does the CPU supports vector constant rotate instructions?
  static constexpr bool supports_vector_constant_rotates(int shift) {
    return false;
  }

  // Does the CPU supports vector unsigned comparison instructions?
  static constexpr bool supports_vector_comparison_unsigned(int vlen, BasicType bt) {
    return false;
  }

  // Some microarchitectures have mask registers used on vectors
  static constexpr bool has_predicated_vectors(void) {
    return false;
  }

  // true means we have fast l2f conversion
  // false means that conversion is done by runtime call
  static const bool convL2FSupported(void) {
    // fcfids can do the conversion (>= Power7).
    // fcfid + frsp showed rounding problem when result should be 0x3f800001.
    return VM_Version::has_fcfids();
  }

  // Implements a variant of EncodeISOArrayNode that encode ASCII only
  static const bool supports_encode_ascii_array = true;

  // Some architecture needs a helper to check for alltrue vector
  static constexpr bool vectortest_needs_second_argument(bool is_alltrue, bool is_predicate) {
    return false;
  }

  // BoolTest mask for vector test intrinsics
  static constexpr BoolTest::mask vectortest_mask(bool is_alltrue, bool is_predicate, int vlen) {
    return BoolTest::illegal;
  }

  // Returns pre-selection estimated size of a vector operation.
  static int vector_op_pre_select_sz_estimate(int vopc, BasicType ety, int vlen) {
    switch(vopc) {
      default: return 0;
      case Op_RoundVF: // fall through
      case Op_RoundVD: {
        return 30;
      }
    }
  }
  // Returns pre-selection estimated size of a scalar operation.
  static int scalar_op_pre_select_sz_estimate(int vopc, BasicType ety) {
    switch(vopc) {
      default: return 0;
      case Op_RoundF: // fall through
      case Op_RoundD: {
        return 30;
      }
    }
  }

#endif // CPU_PPC_MATCHER_PPC_HPP