view src/share/vm/opto/output.hpp @ 10905:f57189b7648d

8257192: Integrate AArch64 JIT port into 8u 7009641: Don't fail VM when CodeCache is full 8073108: [AArch64] Use x86 and SPARC CPU instructions for GHASH acceleration 8130309: Need to bailout cleanly if creation of stubs fails when codecache is out of space (AArch64 changes) 8131779: AARCH64: add Montgomery multiply intrinsic 8132875: AArch64: Fix error introduced into AArch64 CodeCache by commit for 8130309 8135018: AARCH64: Missing memory barriers for CMS collector 8145320: Create unsafe_arraycopy and generic_arraycopy for AArch64 8148328: aarch64: redundant lsr instructions in stub code. 8148783: aarch64: SEGV running SpecJBB2013 8148948: aarch64: generate_copy_longs calls align() incorrectly 8149080: AArch64: Recognise disjoint array copy in stub code 8149365: aarch64: memory copy does not prefetch on backwards copy 8149907: aarch64: use load/store pair instructions in call_stub 8150038: aarch64: make use of CBZ and CBNZ when comparing narrow pointer with zero 8150045: arraycopy causes segfaults in SATB during garbage collection 8150082: aarch64: optimise small array copy 8150229: aarch64: pipeline class for several instructions is not set correctly 8150313: aarch64: optimise array copy using SIMD instructions 8150394: aarch64: add support for 8.1 LSE CAS instructions 8150652: Remove unused code in AArch64 back end 8151340: aarch64: prefetch the destination word for write prior to ldxr/stxr loops. 8151502: optimize pd_disjoint_words and pd_conjoint_words 8151775: aarch64: add support for 8.1 LSE atomic operations 8152537: aarch64: Make use of CBZ and CBNZ when comparing unsigned values with zero. 8152840: aarch64: improve _unsafe_arraycopy stub routine 8153172: aarch64: hotspot crashes after the 8.1 LSE patch is merged 8153713: aarch64: improve short array clearing using store pair 8153797: aarch64: Add Arrays.fill stub code 8154413: AArch64: Better byte behaviour 8154537: AArch64: some integer rotate instructions are never emitted 8154739: AArch64: TemplateTable::fast_xaccess loads in wrong mode 8155015: Aarch64: bad assert in spill generation code 8155100: AArch64: Relax alignment requirement for byte_map_base 8155612: Aarch64: vector nodes need to support misaligned offset 8155617: aarch64: ClearArray does not use DC ZVA 8155627: Enable SA on AArch64 8155653: TestVectorUnalignedOffset.java not pushed with 8155612 8156731: aarch64: java/util/Arrays/Correct.java fails due to _generic_arraycopy stub routine 8157841: aarch64: prefetch ignores cache line size 8157906: aarch64: some more integer rotate instructions are never emitted 8158913: aarch64: SEGV running Spark terasort 8159052: aarch64: optimise unaligned copies in pd_disjoint_words and pd_conjoint_words 8159063: aarch64: optimise unaligned array copy long 8160748: [AArch64] Inconsistent types for ideal_reg 8161072: AArch64: jtreg compiler/uncommontrap/TestDeoptOOM failure 8161190: AArch64: Fix overflow in immediate cmp instruction 8164113: AArch64: follow-up the fix for 8161598 8165673: AArch64: Fix JNI floating point argument handling 8167200: AArch64: Broken stack pointer adjustment in interpreter 8167421: AArch64: in one core system, fatal error: Illegal threadstate encountered 8167595: AArch64: SEGV in stub code cipherBlockChaining_decryptAESCrypt 8168699: Validate special case invocations [AArch64 support] 8168888: Port 8160591: Improve internal array handling to AArch64. 8170100: AArch64: Crash in C1-compiled code accessing References 8170188: jtreg test compiler/types/TestMeetIncompatibleInterfaceArrays.java causes JVM crash 8170873: PPC64/aarch64: Poor StrictMath performance due to non-optimized compilation 8171537: aarch64: compiler/c1/Test6849574.java generates guarantee failure in C1 8172881: AArch64: assertion failure: the int pressure is incorrect 8173472: AArch64: C1 comparisons with null only use 32-bit instructions 8176100: [AArch64] [REDO][REDO] G1 Needs pre barrier on dereference of weak JNI handles 8177661: Correct ad rule output register types from iRegX to iRegXNoSp 8179954: AArch64: C1 and C2 volatile accesses are not sequentially consistent 8182581: aarch64: fix for crash caused by earlyret of compiled method 8183925: [AArch64] Decouple crash protection from watcher thread 8186325: AArch64: jtreg test hotspot/test/gc/g1/TestJNIWeakG1/TestJNIWeakG1.java SEGV 8187224: aarch64: some inconsistency between aarch64_ad.m4 and aarch64.ad 8189170: [AArch64] Add option to disable stack overflow checking in primordial thread for use with JNI_CreateJavaJVM 8193133: Assertion failure because 0xDEADDEAD can be in-heap 8195685: AArch64 port of 8174962: Better interface invocations 8195859: AArch64: vtableStubs gtest fails after 8174962 8196136: AArch64: Correct register use in patch for JDK-8194686 8196221: AArch64: Mistake in committed patch for JDK-8195859 8199712: [AArch64] Flight Recorder 8203481: Incorrect constraint for unextended_sp in frame:safe_for_sender 8203699: java/lang/invoke/SpecialInterfaceCall fails with SIGILL on aarch64 8205421: AARCH64: StubCodeMark should be placed after alignment 8206163: AArch64: incorrect code generation for StoreCM 8207345: Trampoline generation code reads from uninitialized memory 8207838: AArch64: Float registers incorrectly restored in JNI call 8209413: AArch64: NPE in clhsdb jstack command 8209414: [AArch64] method handle invocation does not respect JVMTI interp_only mode 8209415: Fix JVMTI test failure HS202 8209420: Track membars for volatile accesses so they can be properly optimized 8209835: Aarch64: elide barriers on all volatile operations 8210425: [AArch64] sharedRuntimeTrig/sharedRuntimeTrans compiled without optimization 8211064: [AArch64] Interpreter and c1 don't correctly handle jboolean results in native calls 8211233: MemBarNode::trailing_membar() and MemBarNode::leading_membar() need to handle dying subgraphs better 8213134: AArch64: vector shift failed with MaxVectorSize=8 8213419: [AArch64] C2 may hang in MulLNode::Ideal()/MulINode::Ideal() with gcc 8.2.1 8214857: "bad trailing membar" assert failure at memnode.cpp:3220 8215951: AArch64: jtreg test vmTestbase/nsk/jvmti/PopFrame/popframe005 segfaults 8215961: jdk/jfr/event/os/TestCPUInformation.java fails on AArch64 8216350: AArch64: monitor unlock fast path not called 8216989: CardTableBarrierSetAssembler::gen_write_ref_array_post_barrier() does not check for zero length on AARCH64 8217368: AArch64: C2 recursive stack locking optimisation not triggered 8218185: aarch64: missing LoadStore barrier in TemplateTable::putfield_or_static 8219011: Implement MacroAssembler::warn method on AArch64 8219635: aarch64: missing LoadStore barrier in TemplateTable::fast_storefield 8221220: AArch64: Add StoreStore membar explicitly for Volatile Writes in TemplateTable 8221658: aarch64: add necessary predicate for ubfx patterns 8224671: AArch64: mauve System.arraycopy test failure 8224828: aarch64: rflags is not correct after safepoint poll 8224851: AArch64: fix warnings and errors with Clang and GCC 8.3 8224880: AArch64: java/javac error with AllocatePrefetchDistance 8228400: Remove built-in AArch64 simulator 8228406: Superfluous change in chaitin.hpp 8228593: Revert explicit JDK 7 support additions 8228716: Revert InstanceKlass::print_on debug additions 8228718: Revert incorrect backport of JDK-8129757 to 8-aarch64 8228725: AArch64: Purge method call format support 8228747: Revert "unused" attribute from test_arraycopy_func 8228767: Revert ResourceMark additions 8228770: Revert development hsdis changes 8229123: Revert build fixes for aarch64/zero 8229124: Revert disassembler.cpp changes 8229145: Revert TemplateTable::bytecode() visibility change 8233839: aarch64: missing memory barrier in NewObjectArrayStub and NewTypeArrayStub 8237512: AArch64: aarch64TestHook leaks a BufferBlob 8246482: Build failures with +JFR -PCH 8247979: aarch64: missing side effect of killing flags for clearArray_reg_reg 8248219: aarch64: missing memory barrier in fast_storefield and fast_accessfield Reviewed-by: shade, aph
author andrew
date Mon, 01 Feb 2021 03:48:36 +0000
parents ddce0b7cee93
children f79e943d15a7
line wrap: on
line source

/*
 * Copyright (c) 2000, 2015, 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 SHARE_VM_OPTO_OUTPUT_HPP
#define SHARE_VM_OPTO_OUTPUT_HPP

#include "opto/block.hpp"
#include "opto/node.hpp"
#if defined AD_MD_HPP
# include AD_MD_HPP
#elif defined TARGET_ARCH_MODEL_x86_32
# include "adfiles/ad_x86_32.hpp"
#elif defined TARGET_ARCH_MODEL_x86_64
# include "adfiles/ad_x86_64.hpp"
#elif defined TARGET_ARCH_MODEL_aarch64
# include "adfiles/ad_aarch64.hpp"
#elif defined TARGET_ARCH_MODEL_sparc
# include "adfiles/ad_sparc.hpp"
#elif defined TARGET_ARCH_MODEL_zero
# include "adfiles/ad_zero.hpp"
#elif defined TARGET_ARCH_MODEL_ppc_64
# include "adfiles/ad_ppc_64.hpp"
#endif

class Arena;
class Bundle;
class Block;
class Block_Array;
class Node;
class Node_Array;
class Node_List;
class PhaseCFG;
class PhaseChaitin;
class Pipeline_Use_Element;
class Pipeline_Use;

#ifndef PRODUCT
#define DEBUG_ARG(x) , x
#else
#define DEBUG_ARG(x)
#endif

// Define the initial sizes for allocation of the resizable code buffer
enum {
  initial_code_capacity  =  16 * 1024,
  initial_stub_capacity  =   4 * 1024,
  initial_const_capacity =   4 * 1024,
  initial_locs_capacity  =   3 * 1024
};

//------------------------------Scheduling----------------------------------
// This class contains all the information necessary to implement instruction
// scheduling and bundling.
class Scheduling {

private:
  // Arena to use
  Arena *_arena;

  // Control-Flow Graph info
  PhaseCFG *_cfg;

  // Register Allocation info
  PhaseRegAlloc *_regalloc;

  // Number of nodes in the method
  uint _node_bundling_limit;

  // List of scheduled nodes. Generated in reverse order
  Node_List _scheduled;

  // List of nodes currently available for choosing for scheduling
  Node_List _available;

  // For each instruction beginning a bundle, the number of following
  // nodes to be bundled with it.
  Bundle *_node_bundling_base;

  // Mapping from register to Node
  Node_List _reg_node;

  // Free list for pinch nodes.
  Node_List _pinch_free_list;

  // Latency from the beginning of the containing basic block (base 1)
  // for each node.
  unsigned short *_node_latency;

  // Number of uses of this node within the containing basic block.
  short *_uses;

  // Schedulable portion of current block.  Skips Region/Phi/CreateEx up
  // front, branch+proj at end.  Also skips Catch/CProj (same as
  // branch-at-end), plus just-prior exception-throwing call.
  uint _bb_start, _bb_end;

  // Latency from the end of the basic block as scheduled
  unsigned short *_current_latency;

  // Remember the next node
  Node *_next_node;

  // Use this for an unconditional branch delay slot
  Node *_unconditional_delay_slot;

  // Pointer to a Nop
  MachNopNode *_nop;

  // Length of the current bundle, in instructions
  uint _bundle_instr_count;

  // Current Cycle number, for computing latencies and bundling
  uint _bundle_cycle_number;

  // Bundle information
  Pipeline_Use_Element _bundle_use_elements[resource_count];
  Pipeline_Use         _bundle_use;

  // Dump the available list
  void dump_available() const;

public:
  Scheduling(Arena *arena, Compile &compile);

  // Destructor
  NOT_PRODUCT( ~Scheduling(); )

  // Step ahead "i" cycles
  void step(uint i);

  // Step ahead 1 cycle, and clear the bundle state (for example,
  // at a branch target)
  void step_and_clear();

  Bundle* node_bundling(const Node *n) {
    assert(valid_bundle_info(n), "oob");
    return (&_node_bundling_base[n->_idx]);
  }

  bool valid_bundle_info(const Node *n) const {
    return (_node_bundling_limit > n->_idx);
  }

  bool starts_bundle(const Node *n) const {
    return (_node_bundling_limit > n->_idx && _node_bundling_base[n->_idx].starts_bundle());
  }

  // Do the scheduling
  void DoScheduling();

  // Compute the local latencies walking forward over the list of
  // nodes for a basic block
  void ComputeLocalLatenciesForward(const Block *bb);

  // Compute the register antidependencies within a basic block
  void ComputeRegisterAntidependencies(Block *bb);
  void verify_do_def( Node *n, OptoReg::Name def, const char *msg );
  void verify_good_schedule( Block *b, const char *msg );
  void anti_do_def( Block *b, Node *def, OptoReg::Name def_reg, int is_def );
  void anti_do_use( Block *b, Node *use, OptoReg::Name use_reg );

  // Add a node to the current bundle
  void AddNodeToBundle(Node *n, const Block *bb);

  // Add a node to the list of available nodes
  void AddNodeToAvailableList(Node *n);

  // Compute the local use count for the nodes in a block, and compute
  // the list of instructions with no uses in the block as available
  void ComputeUseCount(const Block *bb);

  // Choose an instruction from the available list to add to the bundle
  Node * ChooseNodeToBundle();

  // See if this Node fits into the currently accumulating bundle
  bool NodeFitsInBundle(Node *n);

  // Decrement the use count for a node
 void DecrementUseCounts(Node *n, const Block *bb);

  // Garbage collect pinch nodes for reuse by other blocks.
  void garbage_collect_pinch_nodes();
  // Clean up a pinch node for reuse (helper for above).
  void cleanup_pinch( Node *pinch );

  // Information for statistics gathering
#ifndef PRODUCT
private:
  // Gather information on size of nops relative to total
  uint _branches, _unconditional_delays;

  static uint _total_nop_size, _total_method_size;
  static uint _total_branches, _total_unconditional_delays;
  static uint _total_instructions_per_bundle[Pipeline::_max_instrs_per_cycle+1];

public:
  static void print_statistics();

  static void increment_instructions_per_bundle(uint i) {
    _total_instructions_per_bundle[i]++;
  }

  static void increment_nop_size(uint s) {
    _total_nop_size += s;
  }

  static void increment_method_size(uint s) {
    _total_method_size += s;
  }
#endif

};

#endif // SHARE_VM_OPTO_OUTPUT_HPP