/* * Copyright (c) 1998, 2026, 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_OPTO_LOOPNODE_HPP #define SHARE_OPTO_LOOPNODE_HPP #include "opto/cfgnode.hpp" #include "opto/multnode.hpp" #include "opto/phaseX.hpp" #include "opto/predicates.hpp" #include "opto/subnode.hpp" #include "opto/type.hpp" #include "utilities/checkedCast.hpp" class CmpNode; class BaseCountedLoopEndNode; class CountedLoopNode; class IdealLoopTree; class LoopNode; class Node; class OuterStripMinedLoopEndNode; class PredicateBlock; class PathFrequency; class PhaseIdealLoop; class LoopSelector; class ReachabilityFenceNode; class UnswitchedLoopSelector; class VectorSet; class VSharedData; class Invariance; struct small_cache; // // I D E A L I Z E D L O O P S // // Idealized loops are the set of loops I perform more interesting // transformations on, beyond simple hoisting. //------------------------------LoopNode--------------------------------------- // Simple loop header. Fall in path on left, loop-back path on right. class LoopNode : public RegionNode { // Size is bigger to hold the flags. However, the flags do not change // the semantics so it does not appear in the hash & cmp functions. virtual uint size_of() const { return sizeof(*this); } protected: uint _loop_flags; // Names for flag bitfields enum { Normal=0, Pre=1, Main=2, Post=3, PreMainPostFlagsMask=3, MainHasNoPreLoop = 1<<2, HasExactTripCount = 1<<3, InnerLoop = 1<<4, PartialPeelLoop = 1<<5, PartialPeelFailed = 1<<6, WasSlpAnalyzed = 1<<7, PassedSlpAnalysis = 1<<8, DoUnrollOnly = 1<<9, VectorizedLoop = 1<<10, HasAtomicPostLoop = 1<<11, StripMined = 1<<12, SubwordLoop = 1<<13, ProfileTripFailed = 1<<14, LoopNestInnerLoop = 1<<15, LoopNestLongOuterLoop = 1<<16, MultiversionFastLoop = 1<<17, MultiversionSlowLoop = 2<<17, MultiversionDelayedSlowLoop = 3<<17, MultiversionFlagsMask = 3<<17, }; char _unswitch_count; enum { _unswitch_max=3 }; // Expected trip count from profile data float _profile_trip_cnt; public: // Names for edge indices enum { Self=0, EntryControl, LoopBackControl }; bool is_inner_loop() const { return _loop_flags & InnerLoop; } void set_inner_loop() { _loop_flags |= InnerLoop; } bool is_vectorized_loop() const { return _loop_flags & VectorizedLoop; } bool is_partial_peel_loop() const { return _loop_flags & PartialPeelLoop; } void set_partial_peel_loop() { _loop_flags |= PartialPeelLoop; } bool partial_peel_has_failed() const { return _loop_flags & PartialPeelFailed; } bool is_strip_mined() const { return _loop_flags & StripMined; } bool is_profile_trip_failed() const { return _loop_flags & ProfileTripFailed; } bool is_subword_loop() const { return _loop_flags & SubwordLoop; } bool is_loop_nest_inner_loop() const { return _loop_flags & LoopNestInnerLoop; } bool is_loop_nest_outer_loop() const { return _loop_flags & LoopNestLongOuterLoop; } void mark_partial_peel_failed() { _loop_flags |= PartialPeelFailed; } void mark_was_slp() { _loop_flags |= WasSlpAnalyzed; } void mark_passed_slp() { _loop_flags |= PassedSlpAnalysis; } void mark_do_unroll_only() { _loop_flags |= DoUnrollOnly; } void mark_loop_vectorized() { _loop_flags |= VectorizedLoop; } void mark_has_atomic_post_loop() { _loop_flags |= HasAtomicPostLoop; } void mark_strip_mined() { _loop_flags |= StripMined; } void clear_strip_mined() { _loop_flags &= ~StripMined; } void mark_profile_trip_failed() { _loop_flags |= ProfileTripFailed; } void mark_subword_loop() { _loop_flags |= SubwordLoop; } void mark_loop_nest_inner_loop() { _loop_flags |= LoopNestInnerLoop; } void mark_loop_nest_outer_loop() { _loop_flags |= LoopNestLongOuterLoop; } int unswitch_max() { return _unswitch_max; } int unswitch_count() { return _unswitch_count; } void set_unswitch_count(int val) { assert (val <= unswitch_max(), "too many unswitches"); _unswitch_count = val; } void set_profile_trip_cnt(float ptc) { _profile_trip_cnt = ptc; } float profile_trip_cnt() { return _profile_trip_cnt; } #ifndef PRODUCT uint _stress_peeling_attempts = 0; #endif LoopNode(Node *entry, Node *backedge) : RegionNode(3), _loop_flags(0), _unswitch_count(0), _profile_trip_cnt(COUNT_UNKNOWN) { init_class_id(Class_Loop); init_req(EntryControl, entry); init_req(LoopBackControl, backedge); } virtual Node *Ideal(PhaseGVN *phase, bool can_reshape); virtual int Opcode() const; bool can_be_counted_loop(PhaseValues* phase) const { return req() == 3 && in(0) != nullptr && in(1) != nullptr && phase->type(in(1)) != Type::TOP && in(2) != nullptr && phase->type(in(2)) != Type::TOP; } bool is_valid_counted_loop(BasicType bt) const; #ifndef PRODUCT virtual void dump_spec(outputStream *st) const; #endif void verify_strip_mined(int expect_skeleton) const NOT_DEBUG_RETURN; virtual LoopNode* skip_strip_mined(int expect_skeleton = 1) { return this; } virtual IfTrueNode* outer_loop_tail() const { ShouldNotReachHere(); return nullptr; } virtual OuterStripMinedLoopEndNode* outer_loop_end() const { ShouldNotReachHere(); return nullptr; } virtual IfFalseNode* outer_loop_exit() const { ShouldNotReachHere(); return nullptr; } virtual SafePointNode* outer_safepoint() const { ShouldNotReachHere(); return nullptr; } }; //------------------------------Counted Loops---------------------------------- // Counted loops are all trip-counted loops, with exactly 1 trip-counter exit // path (and maybe some other exit paths). The trip-counter exit is always // last in the loop. The trip-counter have to stride by a constant; // the exit value is also loop invariant. // CountedLoopNodes and CountedLoopEndNodes come in matched pairs. The // CountedLoopNode has the incoming loop control and the loop-back-control // which is always the IfTrue before the matching CountedLoopEndNode. The // CountedLoopEndNode has an incoming control (possibly not the // CountedLoopNode if there is control flow in the loop), the post-increment // trip-counter value, and the limit. The trip-counter value is always of // the form (Op old-trip-counter stride). The old-trip-counter is produced // by a Phi connected to the CountedLoopNode. The stride is constant. // The Op is any commutable opcode, including Add, Mul, Xor. The // CountedLoopEndNode also takes in the loop-invariant limit value. // From a CountedLoopNode I can reach the matching CountedLoopEndNode via the // loop-back control. From CountedLoopEndNodes I can reach CountedLoopNodes // via the old-trip-counter from the Op node. //------------------------------CountedLoopNode-------------------------------- // CountedLoopNodes head simple counted loops. CountedLoopNodes have as // inputs the incoming loop-start control and the loop-back control, so they // act like RegionNodes. They also take in the initial trip counter, the // loop-invariant stride and the loop-invariant limit value. CountedLoopNodes // produce a loop-body control and the trip counter value. Since // CountedLoopNodes behave like RegionNodes I still have a standard CFG model. class BaseCountedLoopNode : public LoopNode { public: BaseCountedLoopNode(Node *entry, Node *backedge) : LoopNode(entry, backedge) { } Node *init_control() const { return in(EntryControl); } Node *back_control() const { return in(LoopBackControl); } Node* init_trip() const; Node* stride() const; bool stride_is_con() const; Node* limit() const; Node* incr() const; Node* phi() const; BaseCountedLoopEndNode* loopexit_or_null() const; BaseCountedLoopEndNode* loopexit() const; virtual BasicType bt() const = 0; jlong stride_con() const; static BaseCountedLoopNode* make(Node* entry, Node* backedge, BasicType bt); virtual void set_trip_count(julong tc) = 0; virtual julong trip_count() const = 0; bool has_exact_trip_count() const { return (_loop_flags & HasExactTripCount) != 0; } void set_exact_trip_count(julong tc) { set_trip_count(tc); _loop_flags |= HasExactTripCount; } void set_nonexact_trip_count() { _loop_flags &= ~HasExactTripCount; } }; class CountedLoopNode : public BaseCountedLoopNode { // Size is bigger to hold _main_idx. However, _main_idx does not change // the semantics so it does not appear in the hash & cmp functions. virtual uint size_of() const { return sizeof(*this); } // For Pre- and Post-loops during debugging ONLY, this holds the index of // the Main CountedLoop. Used to assert that we understand the graph shape. node_idx_t _main_idx; // Known trip count calculated by compute_exact_trip_count() uint _trip_count; // Log2 of original loop bodies in unrolled loop int _unrolled_count_log2; // Node count prior to last unrolling - used to decide if // unroll,optimize,unroll,optimize,... is making progress int _node_count_before_unroll; // If slp analysis is performed we record the maximum // vector mapped unroll factor here int _slp_maximum_unroll_factor; public: CountedLoopNode(Node *entry, Node *backedge) : BaseCountedLoopNode(entry, backedge), _main_idx(0), _trip_count(max_juint), _unrolled_count_log2(0), _node_count_before_unroll(0), _slp_maximum_unroll_factor(0) { init_class_id(Class_CountedLoop); // Initialize _trip_count to the largest possible value. // Will be reset (lower) if the loop's trip count is known. } virtual int Opcode() const; virtual Node *Ideal(PhaseGVN *phase, bool can_reshape); CountedLoopEndNode* loopexit_or_null() const { return (CountedLoopEndNode*) BaseCountedLoopNode::loopexit_or_null(); } CountedLoopEndNode* loopexit() const { return (CountedLoopEndNode*) BaseCountedLoopNode::loopexit(); } int stride_con() const; // A 'main' loop has a pre-loop and a post-loop. The 'main' loop // can run short a few iterations and may start a few iterations in. // It will be RCE'd and unrolled and aligned. // A following 'post' loop will run any remaining iterations. Used // during Range Check Elimination, the 'post' loop will do any final // iterations with full checks. Also used by Loop Unrolling, where // the 'post' loop will do any epilog iterations needed. Basically, // a 'post' loop can not profitably be further unrolled or RCE'd. // A preceding 'pre' loop will run at least 1 iteration (to do peeling), // it may do under-flow checks for RCE and may do alignment iterations // so the following main loop 'knows' that it is striding down cache // lines. // A 'main' loop that is ONLY unrolled or peeled, never RCE'd or // Aligned, may be missing it's pre-loop. bool is_normal_loop () const { return (_loop_flags&PreMainPostFlagsMask) == Normal; } bool is_pre_loop () const { return (_loop_flags&PreMainPostFlagsMask) == Pre; } bool is_main_loop () const { return (_loop_flags&PreMainPostFlagsMask) == Main; } bool is_post_loop () const { return (_loop_flags&PreMainPostFlagsMask) == Post; } bool was_slp_analyzed () const { return (_loop_flags&WasSlpAnalyzed) == WasSlpAnalyzed; } bool has_passed_slp () const { return (_loop_flags&PassedSlpAnalysis) == PassedSlpAnalysis; } bool is_unroll_only () const { return (_loop_flags&DoUnrollOnly) == DoUnrollOnly; } bool is_main_no_pre_loop() const { return _loop_flags & MainHasNoPreLoop; } bool has_atomic_post_loop () const { return (_loop_flags & HasAtomicPostLoop) == HasAtomicPostLoop; } void set_main_no_pre_loop() { _loop_flags |= MainHasNoPreLoop; } IfNode* find_multiversion_if_from_multiversion_fast_main_loop(); int main_idx() const { return _main_idx; } void set_trip_count(julong tc) { assert(tc < max_juint, "Cannot set trip count to max_juint"); _trip_count = checked_cast(tc); } julong trip_count() const { return _trip_count; } void set_pre_loop (CountedLoopNode *main) { assert(is_normal_loop(),""); _loop_flags |= Pre ; _main_idx = main->_idx; } void set_main_loop ( ) { assert(is_normal_loop(),""); _loop_flags |= Main; } void set_post_loop (CountedLoopNode *main) { assert(is_normal_loop(),""); _loop_flags |= Post; _main_idx = main->_idx; } void set_normal_loop( ) { _loop_flags &= ~PreMainPostFlagsMask; } void set_notpassed_slp() { _loop_flags &= ~PassedSlpAnalysis; } void double_unrolled_count() { _unrolled_count_log2++; } int unrolled_count() { return 1 << MIN2(_unrolled_count_log2, BitsPerInt-3); } void set_node_count_before_unroll(int ct) { _node_count_before_unroll = ct; } int node_count_before_unroll() { return _node_count_before_unroll; } void set_slp_max_unroll(int unroll_factor) { _slp_maximum_unroll_factor = unroll_factor; } int slp_max_unroll() const { return _slp_maximum_unroll_factor; } // Multiversioning allows us to duplicate a CountedLoop, and have two versions, and the multiversion_if // decides which one is taken: // (1) fast_loop: We enter this loop by default, by default the multiversion_if has its condition set to // "true", guarded by a OpaqueMultiversioning. If we want to make a speculative assumption // for an optimization, we can add the runtime-check to the multiversion_if, and if the // assumption fails we take the slow_loop instead, where we do not make the same speculative // assumption. // We call it the "fast_loop" because it has more optimizations, enabled by the speculative // runtime-checks at the multiversion_if, and we expect the fast_loop to execute faster. // (2) slow_loop: By default, it is not taken, until a runtime-check is added to the multiversion_if while // optimizing the fast_looop. If such a runtime-check is never added, then after loop-opts // the multiversion_if constant folds to true, and the slow_loop is folded away. To save // compile time, we delay the optimization of the slow_loop until a runtime-check is added // to the multiversion_if, at which point we resume optimizations for the slow_loop. // We call it the "slow_loop" because it has fewer optimizations, since this is the fall-back // loop where we do not make any of the speculative assumptions we make for the fast_loop. // Hence, we expect the slow_loop to execute slower. bool is_multiversion() const { return (_loop_flags & MultiversionFlagsMask) != Normal; } bool is_multiversion_fast_loop() const { return (_loop_flags & MultiversionFlagsMask) == MultiversionFastLoop; } bool is_multiversion_slow_loop() const { return (_loop_flags & MultiversionFlagsMask) == MultiversionSlowLoop; } bool is_multiversion_delayed_slow_loop() const { return (_loop_flags & MultiversionFlagsMask) == MultiversionDelayedSlowLoop; } void set_multiversion_fast_loop() { assert(!is_multiversion(), ""); _loop_flags |= MultiversionFastLoop; } void set_multiversion_slow_loop() { assert(!is_multiversion(), ""); _loop_flags |= MultiversionSlowLoop; } void set_multiversion_delayed_slow_loop() { assert(!is_multiversion(), ""); _loop_flags |= MultiversionDelayedSlowLoop; } void set_no_multiversion() { assert( is_multiversion(), ""); _loop_flags &= ~MultiversionFlagsMask; } virtual LoopNode* skip_strip_mined(int expect_skeleton = 1); OuterStripMinedLoopNode* outer_loop() const; virtual IfTrueNode* outer_loop_tail() const; virtual OuterStripMinedLoopEndNode* outer_loop_end() const; virtual IfFalseNode* outer_loop_exit() const; virtual SafePointNode* outer_safepoint() const; Node* skip_assertion_predicates_with_halt(); virtual BasicType bt() const { return T_INT; } Node* is_canonical_loop_entry(); CountedLoopEndNode* find_pre_loop_end(); Node* uncasted_init_trip(bool uncasted); #ifndef PRODUCT virtual void dump_spec(outputStream *st) const; #endif }; class LongCountedLoopNode : public BaseCountedLoopNode { private: virtual uint size_of() const { return sizeof(*this); } // Known trip count calculated by compute_exact_trip_count() julong _trip_count; public: LongCountedLoopNode(Node *entry, Node *backedge) : BaseCountedLoopNode(entry, backedge), _trip_count(max_julong) { init_class_id(Class_LongCountedLoop); } virtual int Opcode() const; virtual BasicType bt() const { return T_LONG; } void set_trip_count(julong tc) { assert(tc < max_julong, "Cannot set trip count to max_julong"); _trip_count = tc; } julong trip_count() const { return _trip_count; } LongCountedLoopEndNode* loopexit_or_null() const { return (LongCountedLoopEndNode*) BaseCountedLoopNode::loopexit_or_null(); } LongCountedLoopEndNode* loopexit() const { return (LongCountedLoopEndNode*) BaseCountedLoopNode::loopexit(); } }; //------------------------------CountedLoopEndNode----------------------------- // CountedLoopEndNodes end simple trip counted loops. They act much like // IfNodes. class BaseCountedLoopEndNode : public IfNode { public: enum { TestControl, TestValue }; BaseCountedLoopEndNode(Node *control, Node *test, float prob, float cnt) : IfNode(control, test, prob, cnt) { init_class_id(Class_BaseCountedLoopEnd); } Node *cmp_node() const { return (in(TestValue)->req() >=2) ? in(TestValue)->in(1) : nullptr; } Node* incr() const { Node* tmp = cmp_node(); return (tmp && tmp->req() == 3) ? tmp->in(1) : nullptr; } Node* limit() const { Node* tmp = cmp_node(); return (tmp && tmp->req() == 3) ? tmp->in(2) : nullptr; } Node* stride() const { Node* tmp = incr(); return (tmp && tmp->req() == 3) ? tmp->in(2) : nullptr; } Node* init_trip() const { Node* tmp = phi(); return (tmp && tmp->req() == 3) ? tmp->in(1) : nullptr; } bool stride_is_con() const { Node *tmp = stride(); return (tmp != nullptr && tmp->is_Con()); } PhiNode* phi() const { Node* tmp = incr(); if (tmp && tmp->req() == 3) { Node* phi = tmp->in(1); if (phi->is_Phi()) { return phi->as_Phi(); } } return nullptr; } BaseCountedLoopNode* loopnode() const { // The CountedLoopNode that goes with this CountedLoopEndNode may // have been optimized out by the IGVN so be cautious with the // pattern matching on the graph PhiNode* iv_phi = phi(); if (iv_phi == nullptr) { return nullptr; } Node* ln = iv_phi->in(0); if (!ln->is_BaseCountedLoop() || ln->as_BaseCountedLoop()->loopexit_or_null() != this) { return nullptr; } if (ln->as_BaseCountedLoop()->bt() != bt()) { return nullptr; } return ln->as_BaseCountedLoop(); } BoolTest::mask test_trip() const { return in(TestValue)->as_Bool()->_test._test; } jlong stride_con() const; virtual BasicType bt() const = 0; static BaseCountedLoopEndNode* make(Node* control, Node* test, float prob, float cnt, BasicType bt); }; class CountedLoopEndNode : public BaseCountedLoopEndNode { public: CountedLoopEndNode(Node *control, Node *test, float prob, float cnt) : BaseCountedLoopEndNode(control, test, prob, cnt) { init_class_id(Class_CountedLoopEnd); } virtual int Opcode() const; CountedLoopNode* loopnode() const { return (CountedLoopNode*) BaseCountedLoopEndNode::loopnode(); } virtual BasicType bt() const { return T_INT; } #ifndef PRODUCT virtual void dump_spec(outputStream *st) const; #endif }; class LongCountedLoopEndNode : public BaseCountedLoopEndNode { public: LongCountedLoopEndNode(Node *control, Node *test, float prob, float cnt) : BaseCountedLoopEndNode(control, test, prob, cnt) { init_class_id(Class_LongCountedLoopEnd); } LongCountedLoopNode* loopnode() const { return (LongCountedLoopNode*) BaseCountedLoopEndNode::loopnode(); } virtual int Opcode() const; virtual BasicType bt() const { return T_LONG; } }; inline BaseCountedLoopEndNode* BaseCountedLoopNode::loopexit_or_null() const { Node* bctrl = back_control(); if (bctrl == nullptr) return nullptr; Node* lexit = bctrl->in(0); if (!lexit->is_BaseCountedLoopEnd()) { return nullptr; } BaseCountedLoopEndNode* result = lexit->as_BaseCountedLoopEnd(); if (result->bt() != bt()) { return nullptr; } return result; } inline BaseCountedLoopEndNode* BaseCountedLoopNode::loopexit() const { BaseCountedLoopEndNode* cle = loopexit_or_null(); assert(cle != nullptr, "loopexit is null"); return cle; } inline Node* BaseCountedLoopNode::init_trip() const { BaseCountedLoopEndNode* cle = loopexit_or_null(); return cle != nullptr ? cle->init_trip() : nullptr; } inline Node* BaseCountedLoopNode::stride() const { BaseCountedLoopEndNode* cle = loopexit_or_null(); return cle != nullptr ? cle->stride() : nullptr; } inline bool BaseCountedLoopNode::stride_is_con() const { BaseCountedLoopEndNode* cle = loopexit_or_null(); return cle != nullptr && cle->stride_is_con(); } inline Node* BaseCountedLoopNode::limit() const { BaseCountedLoopEndNode* cle = loopexit_or_null(); return cle != nullptr ? cle->limit() : nullptr; } inline Node* BaseCountedLoopNode::incr() const { BaseCountedLoopEndNode* cle = loopexit_or_null(); return cle != nullptr ? cle->incr() : nullptr; } inline Node* BaseCountedLoopNode::phi() const { BaseCountedLoopEndNode* cle = loopexit_or_null(); return cle != nullptr ? cle->phi() : nullptr; } inline jlong BaseCountedLoopNode::stride_con() const { BaseCountedLoopEndNode* cle = loopexit_or_null(); return cle != nullptr ? cle->stride_con() : 0; } //------------------------------LoopLimitNode----------------------------- // Counted Loop limit node which represents exact final iterator value: // trip_count = (limit - init_trip + stride - 1)/stride // final_value= trip_count * stride + init_trip. // Use HW instructions to calculate it when it can overflow in integer. // Note, final_value should fit into integer since counted loop has // limit check: limit <= max_int-stride. class LoopLimitNode : public Node { enum { Init=1, Limit=2, Stride=3 }; public: LoopLimitNode( Compile* C, Node *init, Node *limit, Node *stride ) : Node(nullptr,init,limit,stride) { // Put it on the Macro nodes list to optimize during macro nodes expansion. init_flags(Flag_is_macro); C->add_macro_node(this); } virtual int Opcode() const; virtual const Type *bottom_type() const { return TypeInt::INT; } virtual uint ideal_reg() const { return Op_RegI; } virtual const Type* Value(PhaseGVN* phase) const; virtual Node *Ideal(PhaseGVN *phase, bool can_reshape); virtual Node* Identity(PhaseGVN* phase); }; // Support for strip mining class OuterStripMinedLoopNode : public LoopNode { private: void fix_sunk_stores_when_back_to_counted_loop(PhaseIterGVN* igvn, PhaseIdealLoop* iloop) const; void handle_sunk_stores_when_finishing_construction(PhaseIterGVN* igvn); public: OuterStripMinedLoopNode(Compile* C, Node *entry, Node *backedge) : LoopNode(entry, backedge) { init_class_id(Class_OuterStripMinedLoop); init_flags(Flag_is_macro); C->add_macro_node(this); } virtual int Opcode() const; virtual IfTrueNode* outer_loop_tail() const; virtual OuterStripMinedLoopEndNode* outer_loop_end() const; virtual IfFalseNode* outer_loop_exit() const; virtual SafePointNode* outer_safepoint() const; CountedLoopNode* inner_counted_loop() const { return unique_ctrl_out()->as_CountedLoop(); } CountedLoopEndNode* inner_counted_loop_end() const { return inner_counted_loop()->loopexit(); } IfFalseNode* inner_loop_exit() const { return inner_counted_loop_end()->false_proj(); } void adjust_strip_mined_loop(PhaseIterGVN* igvn); void remove_outer_loop_and_safepoint(PhaseIterGVN* igvn) const; void transform_to_counted_loop(PhaseIterGVN* igvn, PhaseIdealLoop* iloop); static Node* register_new_node(Node* node, LoopNode* ctrl, PhaseIterGVN* igvn, PhaseIdealLoop* iloop); Node* register_control(Node* node, Node* loop, Node* idom, PhaseIterGVN* igvn, PhaseIdealLoop* iloop); }; class OuterStripMinedLoopEndNode : public IfNode { public: OuterStripMinedLoopEndNode(Node *control, Node *test, float prob, float cnt) : IfNode(control, test, prob, cnt) { init_class_id(Class_OuterStripMinedLoopEnd); } virtual int Opcode() const; virtual const Type* Value(PhaseGVN* phase) const; virtual Node *Ideal(PhaseGVN *phase, bool can_reshape); bool is_expanded(PhaseGVN *phase) const; }; // -----------------------------IdealLoopTree---------------------------------- class IdealLoopTree : public ResourceObj { public: IdealLoopTree *_parent; // Parent in loop tree IdealLoopTree *_next; // Next sibling in loop tree IdealLoopTree *_child; // First child in loop tree // The head-tail backedge defines the loop. // If a loop has multiple backedges, this is addressed during cleanup where // we peel off the multiple backedges, merging all edges at the bottom and // ensuring that one proper backedge flow into the loop. Node *_head; // Head of loop Node *_tail; // Tail of loop inline Node *tail(); // Handle lazy update of _tail field inline Node *head(); // Handle lazy update of _head field PhaseIdealLoop* _phase; int _local_loop_unroll_limit; int _local_loop_unroll_factor; Node_List _body; // Loop body for inner loops uint16_t _nest; // Nesting depth uint8_t _irreducible:1, // True if irreducible _has_call:1, // True if has call safepoint _has_sfpt:1, // True if has non-call safepoint _rce_candidate:1, // True if candidate for range check elimination _has_range_checks:1, _has_range_checks_computed:1; Node_List* _safepts; // List of safepoints in this loop Node_List* _required_safept; // A inner loop cannot delete these safepts; Node_List* _reachability_fences; // List of reachability fences in this loop bool _allow_optimizations; // Allow loop optimizations IdealLoopTree(PhaseIdealLoop* phase, Node* head, Node* tail); // Is 'l' a member of 'this'? bool is_member(const IdealLoopTree *l) const; // Test for nested membership // Set loop nesting depth. Accumulate has_call bits. int set_nest( uint depth ); // Split out multiple fall-in edges from the loop header. Move them to a // private RegionNode before the loop. This becomes the loop landing pad. void split_fall_in( PhaseIdealLoop *phase, int fall_in_cnt ); // Split out the outermost loop from this shared header. void split_outer_loop( PhaseIdealLoop *phase ); // Merge all the backedges from the shared header into a private Region. // Feed that region as the one backedge to this loop. void merge_many_backedges( PhaseIdealLoop *phase ); // Split shared headers and insert loop landing pads. // Insert a LoopNode to replace the RegionNode. // Returns TRUE if loop tree is structurally changed. bool beautify_loops( PhaseIdealLoop *phase ); // Perform optimization to use the loop predicates for null checks and range checks. // Applies to any loop level (not just the innermost one) bool loop_predication( PhaseIdealLoop *phase); bool can_apply_loop_predication(); // Perform iteration-splitting on inner loops. Split iterations to // avoid range checks or one-shot null checks. Returns false if the // current round of loop opts should stop. bool iteration_split( PhaseIdealLoop *phase, Node_List &old_new ); // Driver for various flavors of iteration splitting. Returns false // if the current round of loop opts should stop. bool iteration_split_impl( PhaseIdealLoop *phase, Node_List &old_new ); // Given dominators, try to find loops with calls that must always be // executed (call dominates loop tail). These loops do not need non-call // safepoints (ncsfpt). void check_safepts(VectorSet &visited, Node_List &stack); // Allpaths backwards scan from loop tail, terminating each path at first safepoint // encountered. void allpaths_check_safepts(VectorSet &visited, Node_List &stack); // Remove safepoints from loop. Optionally keeping one. void remove_safepoints(PhaseIdealLoop* phase, bool keep_one); // Convert to counted loops where possible void counted_loop( PhaseIdealLoop *phase ); // Check for Node being a loop-breaking test Node *is_loop_exit(Node *iff) const; // Return unique loop-exit projection or null if the loop has multiple exits. IfFalseNode* unique_loop_exit_proj_or_null(); // Remove simplistic dead code from loop body void DCE_loop_body(); // Look for loop-exit tests with my 50/50 guesses from the Parsing stage. // Replace with a 1-in-10 exit guess. void adjust_loop_exit_prob( PhaseIdealLoop *phase ); // Return TRUE or FALSE if the loop should never be RCE'd or aligned. // Useful for unrolling loops with NO array accesses. bool policy_peel_only( PhaseIdealLoop *phase ) const; // Return TRUE or FALSE if the loop should be unswitched -- clone // loop with an invariant test bool policy_unswitching( PhaseIdealLoop *phase ) const; // Micro-benchmark spamming. Remove empty loops. bool do_remove_empty_loop( PhaseIdealLoop *phase ); // Convert one-iteration loop into normal code. bool do_one_iteration_loop( PhaseIdealLoop *phase ); // Return TRUE or FALSE if the loop should be peeled or not. Peel if we can // move some loop-invariant test (usually a null-check) before the loop. bool policy_peeling(PhaseIdealLoop *phase); uint estimate_peeling(PhaseIdealLoop *phase); // Return TRUE or FALSE if the loop should be maximally unrolled. Stash any // known trip count in the counted loop node. bool policy_maximally_unroll(PhaseIdealLoop *phase) const; // Return TRUE or FALSE if the loop should be unrolled or not. Apply unroll // if the loop is a counted loop and the loop body is small enough. bool policy_unroll(PhaseIdealLoop *phase); // Loop analyses to map to a maximal superword unrolling for vectorization. void policy_unroll_slp_analysis(CountedLoopNode *cl, PhaseIdealLoop *phase, int future_unroll_ct); // Return TRUE or FALSE if the loop should be range-check-eliminated. // Gather a list of IF tests that are dominated by iteration splitting; // also gather the end of the first split and the start of the 2nd split. bool policy_range_check(PhaseIdealLoop* phase, bool provisional, BasicType bt) const; // Return TRUE if "iff" is a range check. bool is_range_check_if(IfProjNode* if_success_proj, PhaseIdealLoop* phase, Invariance& invar DEBUG_ONLY(COMMA ProjNode* predicate_proj)) const; bool is_range_check_if(IfProjNode* if_success_proj, PhaseIdealLoop* phase, BasicType bt, Node* iv, Node*& range, Node*& offset, jlong& scale) const; // Estimate the number of nodes required when cloning a loop (body). uint est_loop_clone_sz(uint factor) const; // Estimate the number of nodes required when unrolling a loop (body). uint est_loop_unroll_sz(uint factor) const; // Compute loop trip count if possible void compute_trip_count(PhaseIdealLoop* phase, BasicType bt); // Compute loop trip count from profile data float compute_profile_trip_cnt_helper(Node* n); void compute_profile_trip_cnt( PhaseIdealLoop *phase ); // Reassociate invariant expressions. void reassociate_invariants(PhaseIdealLoop *phase); // Reassociate invariant binary expressions. Node* reassociate(Node* n1, PhaseIdealLoop *phase); // Reassociate invariant add, subtract, and compare expressions. Node* reassociate_add_sub_cmp(Node* n1, int inv1_idx, int inv2_idx, PhaseIdealLoop* phase); // Return nonzero index of invariant operand if invariant and variant // are combined with an associative binary. Helper for reassociate_invariants. int find_invariant(Node* n, PhaseIdealLoop *phase); // Return TRUE if "n" is associative. bool is_associative(Node* n, Node* base=nullptr); // Return TRUE if "n" is an associative cmp node. bool is_associative_cmp(Node* n); // Return true if n is invariant bool is_invariant(Node* n) const; // Put loop body on igvn work list void record_for_igvn(); bool is_root() { return _parent == nullptr; } // A proper/reducible loop w/o any (occasional) dead back-edge. bool is_loop() { return !_irreducible && !tail()->is_top(); } bool is_counted() { return is_loop() && _head->is_CountedLoop(); } bool is_innermost() { return is_loop() && _child == nullptr; } void remove_main_post_loops(CountedLoopNode *cl, PhaseIdealLoop *phase); bool compute_has_range_checks() const; bool range_checks_present() { if (!_has_range_checks_computed) { if (compute_has_range_checks()) { _has_range_checks = 1; } _has_range_checks_computed = 1; } return _has_range_checks; } // Return the parent's IdealLoopTree for a strip mined loop which is the outer strip mined loop. // In all other cases, return this. IdealLoopTree* skip_strip_mined() { return _head->as_Loop()->is_strip_mined() ? _parent : this; } // Registers a reachability fence node in the loop. void register_reachability_fence(ReachabilityFenceNode* rf); #ifndef PRODUCT void dump_head(); // Dump loop head only void dump(); // Dump this loop recursively #endif #ifdef ASSERT GrowableArray collect_sorted_children() const; bool verify_tree(IdealLoopTree* loop_verify) const; #endif private: enum { EMPTY_LOOP_SIZE = 7 }; // Number of nodes in an empty loop. // Estimate the number of nodes resulting from control and data flow merge. uint est_loop_flow_merge_sz() const; // Check if the number of residual iterations is large with unroll_cnt. // Return true if the residual iterations are more than 10% of the trip count. bool is_residual_iters_large(int unroll_cnt, CountedLoopNode *cl) const { return (unroll_cnt - 1) * (100.0 / LoopPercentProfileLimit) > cl->profile_trip_cnt(); } void collect_loop_core_nodes(PhaseIdealLoop* phase, Unique_Node_List& wq) const; bool empty_loop_with_data_nodes(PhaseIdealLoop* phase) const; void enqueue_data_nodes(PhaseIdealLoop* phase, Unique_Node_List& empty_loop_nodes, Unique_Node_List& wq) const; bool process_safepoint(PhaseIdealLoop* phase, Unique_Node_List& empty_loop_nodes, Unique_Node_List& wq, Node* sfpt) const; bool empty_loop_candidate(PhaseIdealLoop* phase) const; bool empty_loop_with_extra_nodes_candidate(PhaseIdealLoop* phase) const; }; // -----------------------------PhaseIdealLoop--------------------------------- // Computes the mapping from Nodes to IdealLoopTrees. Organizes IdealLoopTrees // into a loop tree. Drives the loop-based transformations on the ideal graph. class PhaseIdealLoop : public PhaseTransform { friend class IdealLoopTree; friend class SuperWord; friend class AutoNodeBudget; Arena _arena; // For data whose lifetime is a single pass of loop optimizations // Map loop membership for CFG nodes, and ctrl for non-CFG nodes. // // Exception: dead CFG nodes may instead have a ctrl/idom forwarding // installed. See: forward_ctrl Node_List _loop_or_ctrl; // Pre-computed def-use info PhaseIterGVN &_igvn; // Head of loop tree IdealLoopTree* _ltree_root; // Array of pre-order numbers, plus post-visited bit. // ZERO for not pre-visited. EVEN for pre-visited but not post-visited. // ODD for post-visited. Other bits are the pre-order number. uint *_preorders; uint _max_preorder; ReallocMark _nesting; // Safety checks for arena reallocation const PhaseIdealLoop* _verify_me; bool _verify_only; // Allocate _preorders[] array void allocate_preorders() { _max_preorder = C->unique()+8; _preorders = NEW_RESOURCE_ARRAY(uint, _max_preorder); memset(_preorders, 0, sizeof(uint) * _max_preorder); } // Allocate _preorders[] array void reallocate_preorders() { _nesting.check(); // Check if a potential re-allocation in the resource arena is safe if ( _max_preorder < C->unique() ) { _preorders = REALLOC_RESOURCE_ARRAY(_preorders, _max_preorder, C->unique()); _max_preorder = C->unique(); } memset(_preorders, 0, sizeof(uint) * _max_preorder); } // Check to grow _preorders[] array for the case when build_loop_tree_impl() // adds new nodes. void check_grow_preorders( ) { _nesting.check(); // Check if a potential re-allocation in the resource arena is safe if ( _max_preorder < C->unique() ) { uint newsize = _max_preorder<<1; // double size of array _preorders = REALLOC_RESOURCE_ARRAY(_preorders, _max_preorder, newsize); memset(&_preorders[_max_preorder],0,sizeof(uint)*(newsize-_max_preorder)); _max_preorder = newsize; } } // Check for pre-visited. Zero for NOT visited; non-zero for visited. int is_visited( Node *n ) const { return _preorders[n->_idx]; } // Pre-order numbers are written to the Nodes array as low-bit-set values. void set_preorder_visited( Node *n, int pre_order ) { assert( !is_visited( n ), "already set" ); _preorders[n->_idx] = (pre_order<<1); }; // Return pre-order number. int get_preorder( Node *n ) const { assert( is_visited(n), "" ); return _preorders[n->_idx]>>1; } // Check for being post-visited. // Should be previsited already (checked with assert(is_visited(n))). int is_postvisited( Node *n ) const { assert( is_visited(n), "" ); return _preorders[n->_idx]&1; } // Mark as post visited void set_postvisited( Node *n ) { assert( !is_postvisited( n ), "" ); _preorders[n->_idx] |= 1; } public: // Set/get control node out. Set lower bit to distinguish from IdealLoopTree // Returns true if "n" is a data node, false if it's a CFG node. // // Exception: // control nodes that are dead because of "replace_node_and_forward_ctrl" // or have otherwise modified their ctrl state by "forward_ctrl". // They return "true", because they have a ctrl "forwarding" to the other ctrl node they // were replaced with. bool has_ctrl(const Node* n) const { return ((intptr_t)_loop_or_ctrl[n->_idx]) & 1; } private: // clear out dead code after build_loop_late Node_List _deadlist; Node_List _zero_trip_guard_opaque_nodes; Node_List _multiversion_opaque_nodes; // Support for faster execution of get_late_ctrl()/dom_lca() // when a node has many uses and dominator depth is deep. GrowableArray _dom_lca_tags; uint _dom_lca_tags_round; void init_dom_lca_tags(); // Helper for debugging bad dominance relationships bool verify_dominance(Node* n, Node* use, Node* LCA, Node* early); Node* compute_lca_of_uses(Node* n, Node* early, bool verify = false); // Inline wrapper for frequent cases: // 1) only one use // 2) a use is the same as the current LCA passed as 'n1' Node *dom_lca_for_get_late_ctrl( Node *lca, Node *n, Node *tag ) { assert( n->is_CFG(), "" ); // Fast-path null lca if( lca != nullptr && lca != n ) { assert( lca->is_CFG(), "" ); // find LCA of all uses n = dom_lca_for_get_late_ctrl_internal( lca, n, tag ); } return find_non_split_ctrl(n); } Node *dom_lca_for_get_late_ctrl_internal( Node *lca, Node *n, Node *tag ); // Helper function for directing control inputs away from CFG split points. Node *find_non_split_ctrl( Node *ctrl ) const { if (ctrl != nullptr) { if (ctrl->is_MultiBranch()) { ctrl = ctrl->in(0); } assert(ctrl->is_CFG(), "CFG"); } return ctrl; } void cast_incr_before_loop(Node* incr, Node* ctrl, CountedLoopNode* loop); #ifdef ASSERT static void ensure_zero_trip_guard_proj(Node* node, bool is_main_loop); #endif private: static void get_opaque_template_assertion_predicate_nodes(ParsePredicateSuccessProj* parse_predicate_proj, Unique_Node_List& list); void update_main_loop_assertion_predicates(CountedLoopNode* new_main_loop_head, int stride_con_before_unroll); void initialize_assertion_predicates_for_peeled_loop(CountedLoopNode* peeled_loop_head, CountedLoopNode* remaining_loop_head, uint first_node_index_in_cloned_loop_body, const Node_List& old_new); void initialize_assertion_predicates_for_main_loop(CountedLoopNode* pre_loop_head, CountedLoopNode* main_loop_head, uint first_node_index_in_pre_loop_body, uint last_node_index_in_pre_loop_body, DEBUG_ONLY(uint last_node_index_from_backedge_goo COMMA) const Node_List& old_new); void initialize_assertion_predicates_for_post_loop(CountedLoopNode* main_loop_head, CountedLoopNode* post_loop_head, uint first_node_index_in_cloned_loop_body); void create_assertion_predicates_at_loop(CountedLoopNode* source_loop_head, CountedLoopNode* target_loop_head, const NodeInLoopBody& _node_in_loop_body, bool kill_old_template); void create_assertion_predicates_at_main_or_post_loop(CountedLoopNode* source_loop_head, CountedLoopNode* target_loop_head, const NodeInLoopBody& _node_in_loop_body, bool kill_old_template); void rewire_old_target_loop_entry_dependency_to_new_entry(CountedLoopNode* target_loop_head, const Node* old_target_loop_entry, uint node_index_before_new_assertion_predicate_nodes); void log_loop_tree(); public: PhaseIterGVN &igvn() const { return _igvn; } Arena* arena() { return &_arena; }; bool has_node(const Node* n) const { guarantee(n != nullptr, "No Node."); return _loop_or_ctrl[n->_idx] != nullptr; } // check if transform created new nodes that need _ctrl recorded Node *get_late_ctrl( Node *n, Node *early ); Node *get_early_ctrl( Node *n ); Node *get_early_ctrl_for_expensive(Node *n, Node* earliest); void set_early_ctrl(Node* n, bool update_body); void set_subtree_ctrl(Node* n, bool update_body); void set_ctrl( Node *n, Node *ctrl ) { assert( !has_node(n) || has_ctrl(n), "" ); assert( ctrl->in(0), "cannot set dead control node" ); assert( ctrl == find_non_split_ctrl(ctrl), "must set legal crtl" ); _loop_or_ctrl.map(n->_idx, (Node*)((intptr_t)ctrl + 1)); } void set_root_as_ctrl(Node* n) { assert(!has_node(n) || has_ctrl(n), ""); _loop_or_ctrl.map(n->_idx, (Node*)((intptr_t)C->root() + 1)); } // Set control and update loop membership void set_ctrl_and_loop(Node* n, Node* ctrl) { IdealLoopTree* old_loop = get_loop(get_ctrl(n)); IdealLoopTree* new_loop = get_loop(ctrl); if (old_loop != new_loop) { if (old_loop->_child == nullptr) old_loop->_body.yank(n); if (new_loop->_child == nullptr) new_loop->_body.push(n); } set_ctrl(n, ctrl); } // Retrieves the ctrl for a data node i. Node* get_ctrl(const Node* i) { assert(has_node(i) && has_ctrl(i), "must be data node with ctrl"); Node* n = get_ctrl_no_update(i); // We store the found ctrl in the side-table again. In most cases, // this is a no-op, since we just read from _loop_or_ctrl. But in cases // where there was a ctrl forwarding via dead ctrl nodes, this shortens the path. // See: forward_ctrl _loop_or_ctrl.map(i->_idx, (Node*)((intptr_t)n + 1)); assert(has_node(i) && has_ctrl(i), "must still be data node with ctrl"); assert(n == find_non_split_ctrl(n), "must return legal ctrl"); return n; } bool is_dominator(Node* dominator, Node* n); bool is_strict_dominator(Node* dominator, Node* n); // return get_ctrl for a data node and self(n) for a CFG node Node* ctrl_or_self(Node* n) { if (has_ctrl(n)) return get_ctrl(n); else { assert (n->is_CFG(), "must be a CFG node"); return n; } } private: Node* get_ctrl_no_update_helper(const Node* i) const { // We expect only data nodes (which must have a ctrl set), or // dead ctrl nodes that have a ctrl "forwarding". // See: forward_ctrl. assert(has_ctrl(i), "only data nodes or ctrl nodes with ctrl forwarding expected"); return (Node*)(((intptr_t)_loop_or_ctrl[i->_idx]) & ~1); } // Compute the ctrl of node i, jumping over ctrl forwardings. Node* get_ctrl_no_update(const Node* i) const { assert(has_ctrl(i), "only data nodes expected"); Node* n = get_ctrl_no_update_helper(i); if (n->in(0) == nullptr) { // We encountered a dead CFG node. // If everything went right, this dead CFG node should have had a ctrl // forwarding installed, using "forward_ctrl". We now have to jump from // the old (dead) ctrl node to the new (live) ctrl node, in possibly // multiple ctrl forwarding steps. do { n = get_ctrl_no_update_helper(n); } while (n->in(0) == nullptr); n = find_non_split_ctrl(n); } return n; } public: // Check for loop being set // "n" must be a control node. Returns true if "n" is known to be in a loop. bool has_loop( Node *n ) const { assert(!has_node(n) || !has_ctrl(n), ""); return has_node(n); } // Set loop void set_loop( Node *n, IdealLoopTree *loop ) { _loop_or_ctrl.map(n->_idx, (Node*)loop); } // Install a ctrl "forwarding" from an old (dead) control node. // This is a "lazy" update of the "get_ctrl" and "idom" mechanism: // - Install a forwarding from old_node (dead ctrl) to new_node. // - When querying "get_ctrl": jump from data node over possibly // multiple dead ctrl nodes with ctrl forwarding to eventually // reach a live ctrl node. Shorten the path to avoid chasing the // forwarding in the future. // - When querying "idom": from some node get its old idom, which // may be dead but has an idom forwarding to the new and live // idom. Shorten the path to avoid chasing the forwarding in the // future. // Note: while the "idom" information is stored in the "_idom" // side-table, the idom forwarding piggybacks on the ctrl // forwarding on "_loop_or_ctrl". // Using "forward_ctrl" allows us to only edit the entry for the old // dead node now, and we do not have to update all the nodes that had // the old_node as their "get_ctrl" or "idom". We clean up the forwarding // links when we query "get_ctrl" or "idom" for these nodes the next time. void forward_ctrl(Node* old_node, Node* new_node) { assert(!has_ctrl(old_node) && old_node->is_CFG() && old_node->in(0) == nullptr, "must be dead ctrl (CFG) node"); assert(!has_ctrl(new_node) && new_node->is_CFG() && new_node->in(0) != nullptr, "must be live ctrl (CFG) node"); assert(old_node != new_node, "no cycles please"); // Re-use the side array slot for this node to provide the // forwarding pointer. _loop_or_ctrl.map(old_node->_idx, (Node*)((intptr_t)new_node + 1)); assert(has_ctrl(old_node), "must have installed ctrl forwarding"); } // Replace the old ctrl node with a new ctrl node. // - Update the node inputs of all uses. // - Lazily update the ctrl and idom info of all uses, via a ctrl/idom forwarding. void replace_node_and_forward_ctrl(Node* old_node, Node* new_node) { _igvn.replace_node(old_node, new_node); forward_ctrl(old_node, new_node); } void remove_dead_data_node(Node* dead) { assert(dead->outcnt() == 0 && !dead->is_top(), "must be dead"); assert(!dead->is_CFG(), "not a data node"); Node* c = get_ctrl(dead); IdealLoopTree* lpt = get_loop(c); _loop_or_ctrl.map(dead->_idx, nullptr); // This node is useless lpt->_body.yank(dead); igvn().remove_dead_node(dead, PhaseIterGVN::NodeOrigin::Graph); } private: // Place 'n' in some loop nest, where 'n' is a CFG node void build_loop_tree(); int build_loop_tree_impl(Node* n, int pre_order); // Insert loop into the existing loop tree. 'innermost' is a leaf of the // loop tree, not the root. IdealLoopTree *sort( IdealLoopTree *loop, IdealLoopTree *innermost ); #ifdef ASSERT // verify that regions in irreducible loops are marked is_in_irreducible_loop void verify_regions_in_irreducible_loops(); bool is_in_irreducible_loop(RegionNode* region); #endif // Place Data nodes in some loop nest void build_loop_early( VectorSet &visited, Node_List &worklist, Node_Stack &nstack ); void build_loop_late ( VectorSet &visited, Node_List &worklist, Node_Stack &nstack ); void build_loop_late_post_work(Node* n, bool pinned); void build_loop_late_post(Node* n); void verify_strip_mined_scheduling(Node *n, Node* least); // Array of immediate dominance info for each CFG node indexed by node idx private: uint _idom_size; Node **_idom; // Array of immediate dominators uint *_dom_depth; // Used for fast LCA test GrowableArray* _dom_stk; // For recomputation of dom depth LoopOptsMode _mode; // build the loop tree and perform any requested optimizations void build_and_optimize(); // Dominators for the sea of nodes void Dominators(); // Compute the Ideal Node to Loop mapping PhaseIdealLoop(PhaseIterGVN& igvn, LoopOptsMode mode) : PhaseTransform(Ideal_Loop), _arena(mtCompiler, Arena::Tag::tag_idealloop), _loop_or_ctrl(&_arena), _igvn(igvn), _verify_me(nullptr), _verify_only(false), _mode(mode), _nodes_required(UINT_MAX) { assert(mode != LoopOptsVerify, "wrong constructor to verify IdealLoop"); build_and_optimize(); } #ifndef PRODUCT // Verify that verify_me made the same decisions as a fresh run // or only verify that the graph is valid if verify_me is null. PhaseIdealLoop(PhaseIterGVN& igvn, const PhaseIdealLoop* verify_me = nullptr) : PhaseTransform(Ideal_Loop), _arena(mtCompiler, Arena::Tag::tag_idealloop), _loop_or_ctrl(&_arena), _igvn(igvn), _verify_me(verify_me), _verify_only(verify_me == nullptr), _mode(LoopOptsVerify), _nodes_required(UINT_MAX) { DEBUG_ONLY(C->set_phase_verify_ideal_loop();) build_and_optimize(); DEBUG_ONLY(C->reset_phase_verify_ideal_loop();) } #endif Node* insert_convert_node_if_needed(BasicType target, Node* input); Node* idom_no_update(Node* d) const { return idom_no_update(d->_idx); } Node* idom_no_update(uint node_idx) const { assert(node_idx < _idom_size, "oob"); Node* n = _idom[node_idx]; assert(n != nullptr,"Bad immediate dominator info."); while (n->in(0) == nullptr) { // Skip dead CFG nodes // We encountered a dead CFG node. // If everything went right, this dead CFG node should have had an idom // forwarding installed, using "forward_ctrl". We now have to jump from // the old (dead) idom node to the new (live) idom node, in possibly // multiple idom forwarding steps. // Note that we piggyback on "_loop_or_ctrl" to do the forwarding, // since we forward both "get_ctrl" and "idom" from the dead to the // new live ctrl/idom nodes. n = (Node*)(((intptr_t)_loop_or_ctrl[n->_idx]) & ~1); assert(n != nullptr,"Bad immediate dominator info."); } return n; } public: Node* idom(Node* n) const { return idom(n->_idx); } Node* idom(uint node_idx) const { Node* n = idom_no_update(node_idx); // We store the found idom in the side-table again. In most cases, // this is a no-op, since we just read from _idom. But in cases where // there was an idom forwarding via dead idom nodes, this shortens the path. // See: forward_ctrl _idom[node_idx] = n; return n; } uint dom_depth(Node* d) const { guarantee(d != nullptr, "Null dominator info."); guarantee(d->_idx < _idom_size, ""); return _dom_depth[d->_idx]; } void set_idom(Node* d, Node* n, uint dom_depth); // Locally compute IDOM using dom_lca call Node *compute_idom( Node *region ) const; // Recompute dom_depth void recompute_dom_depth(); // Is safept not required by an outer loop? bool is_deleteable_safept(Node* sfpt) const; // Replace parallel induction variable (parallel to trip counter) void replace_parallel_iv(IdealLoopTree *loop); Node *dom_lca( Node *n1, Node *n2 ) const { return find_non_split_ctrl(dom_lca_internal(n1, n2)); } Node *dom_lca_internal( Node *n1, Node *n2 ) const; Node* dominated_node(Node* c1, Node* c2) { assert(is_dominator(c1, c2) || is_dominator(c2, c1), "nodes must be related"); return is_dominator(c1, c2) ? c2 : c1; } // Return control node that's dominated by the 2 others Node* dominated_node(Node* c1, Node* c2, Node* c3) { return dominated_node(c1, dominated_node(c2, c3)); } // Build and verify the loop tree without modifying the graph. This // is useful to verify that all inputs properly dominate their uses. static void verify(PhaseIterGVN& igvn) { #ifdef ASSERT ResourceMark rm; Compile::TracePhase tp(_t_idealLoopVerify); PhaseIdealLoop v(igvn); #endif } // Recommended way to use PhaseIdealLoop. // Run PhaseIdealLoop in some mode and allocates a local scope for memory allocations. static void optimize(PhaseIterGVN &igvn, LoopOptsMode mode) { ResourceMark rm; PhaseIdealLoop v(igvn, mode); Compile* C = Compile::current(); if (!C->failing()) { // Cleanup any modified bits igvn.optimize(); if (C->failing()) { return; } v.log_loop_tree(); } } // True if the method has at least 1 irreducible loop bool _has_irreducible_loops; // Per-Node transform virtual Node* transform(Node* n) { return nullptr; } Node* loop_exit_control(const IdealLoopTree* loop) const; class LoopExitTest { bool _is_valid; const Node* _back_control; const IdealLoopTree* _loop; PhaseIdealLoop* _phase; Node* _cmp; Node* _incr; Node* _limit; BoolTest::mask _mask; float _cl_prob; public: LoopExitTest(const Node* back_control, const IdealLoopTree* loop, PhaseIdealLoop* phase) : _is_valid(false), _back_control(back_control), _loop(loop), _phase(phase), _cmp(nullptr), _incr(nullptr), _limit(nullptr), _mask(BoolTest::illegal), _cl_prob(0.0f) {} void build(); void canonicalize_mask(jlong stride_con); bool is_valid_with_bt(BasicType bt) const { return _is_valid && _cmp != nullptr && _cmp->Opcode() == Op_Cmp(bt); } bool should_include_limit() const { return _mask == BoolTest::le || _mask == BoolTest::ge; } CmpNode* cmp() const { return _cmp->as_Cmp(); } Node* incr() const { return _incr; } Node* limit() const { return _limit; } BoolTest::mask mask() const { return _mask; } float cl_prob() const { return _cl_prob; } }; class LoopIVIncr { bool _is_valid; const Node* _head; const IdealLoopTree* _loop; Node* _incr; Node* _phi_incr; public: LoopIVIncr(const Node* head, const IdealLoopTree* loop) : _is_valid(false), _head(head), _loop(loop), _incr(nullptr), _phi_incr(nullptr) {} void build(Node* old_incr); bool is_valid() const { return _is_valid; } bool is_valid_with_bt(const BasicType bt) const { return _is_valid && _incr->Opcode() == Op_Add(bt); } Node* incr() const { return _incr; } Node* phi_incr() const { return _phi_incr; } }; class LoopIVStride { bool _is_valid; BasicType _iv_bt; Node* _stride_node; Node* _xphi; public: LoopIVStride(BasicType iv_bt) : _is_valid(false), _iv_bt(iv_bt), _stride_node(nullptr), _xphi(nullptr) {} void build(const Node* incr); bool is_valid() const { return _is_valid && _stride_node != nullptr; } Node* stride_node() const { return _stride_node; } Node* xphi() const { return _xphi; } jlong compute_non_zero_stride_con(BoolTest::mask mask, BasicType iv_bt) const; }; static PhiNode* loop_iv_phi(const Node* xphi, const Node* phi_incr, const Node* head); bool try_convert_to_counted_loop(Node* head, IdealLoopTree*& loop, BasicType iv_bt); Node* loop_nest_replace_iv(Node* iv_to_replace, Node* inner_iv, Node* outer_phi, Node* inner_head, BasicType bt); bool create_loop_nest(IdealLoopTree* loop, Node_List &old_new); void add_parse_predicate(Deoptimization::DeoptReason reason, Node* inner_head, IdealLoopTree* loop, SafePointNode* sfpt); SafePointNode* find_safepoint(Node* back_control, const Node* head, const IdealLoopTree* loop); void add_parse_predicates(IdealLoopTree* outer_ilt, LoopNode* inner_head, SafePointNode* cloned_sfpt); IdealLoopTree* insert_outer_loop(IdealLoopTree* loop, LoopNode* outer_l, Node* outer_ift); IdealLoopTree* create_outer_strip_mined_loop(Node* init_control, IdealLoopTree* loop, float cl_prob, float le_fcnt, Node*& entry_control, Node*& iffalse); Node* exact_limit( IdealLoopTree *loop ); // Return a post-walked LoopNode IdealLoopTree* get_loop(const Node* n) const { // Dead nodes have no loop, so return the top level loop instead if (!has_node(n)) return _ltree_root; assert(!has_ctrl(n), ""); return (IdealLoopTree*)_loop_or_ctrl[n->_idx]; } IdealLoopTree* ltree_root() const { return _ltree_root; } // Is 'n' a (nested) member of 'loop'? bool is_member(const IdealLoopTree* loop, const Node* n) const { return loop->is_member(get_loop(n)); } // is the control for 'n' a (nested) member of 'loop'? bool ctrl_is_member(const IdealLoopTree* loop, const Node* n) { return is_member(loop, get_ctrl(n)); } // This is the basic building block of the loop optimizations. It clones an // entire loop body. It makes an old_new loop body mapping; with this // mapping you can find the new-loop equivalent to an old-loop node. All // new-loop nodes are exactly equal to their old-loop counterparts, all // edges are the same. All exits from the old-loop now have a RegionNode // that merges the equivalent new-loop path. This is true even for the // normal "loop-exit" condition. All uses of loop-invariant old-loop values // now come from (one or more) Phis that merge their new-loop equivalents. // Parameter side_by_side_idom: // When side_by_size_idom is null, the dominator tree is constructed for // the clone loop to dominate the original. Used in construction of // pre-main-post loop sequence. // When nonnull, the clone and original are side-by-side, both are // dominated by the passed in side_by_side_idom node. Used in // construction of unswitched loops. enum CloneLoopMode { IgnoreStripMined = 0, // Only clone inner strip mined loop CloneIncludesStripMined = 1, // clone both inner and outer strip mined loops ControlAroundStripMined = 2 // Only clone inner strip mined loop, // result control flow branches // either to inner clone or outer // strip mined loop. }; void clone_loop( IdealLoopTree *loop, Node_List &old_new, int dom_depth, CloneLoopMode mode, Node* side_by_side_idom = nullptr); void clone_loop_handle_data_uses(Node* old, Node_List &old_new, IdealLoopTree* loop, IdealLoopTree* companion_loop, Node_List*& split_if_set, Node_List*& split_bool_set, Node_List*& split_cex_set, Node_List& worklist, uint new_counter, CloneLoopMode mode); void clone_outer_loop(LoopNode* head, CloneLoopMode mode, IdealLoopTree *loop, IdealLoopTree* outer_loop, int dd, Node_List &old_new, Node_List& extra_data_nodes); // If we got the effect of peeling, either by actually peeling or by // making a pre-loop which must execute at least once, we can remove // all loop-invariant dominated tests in the main body. void peeled_dom_test_elim( IdealLoopTree *loop, Node_List &old_new ); // Generate code to do a loop peel for the given loop (and body). // old_new is a temp array. void do_peeling( IdealLoopTree *loop, Node_List &old_new ); // Add pre and post loops around the given loop. These loops are used // during RCE, unrolling and aligning loops. void insert_pre_post_loops( IdealLoopTree *loop, Node_List &old_new, bool peel_only ); // Find the last store in the body of an OuterStripMinedLoop when following memory uses Node *find_last_store_in_outer_loop(Node* store, const IdealLoopTree* outer_loop); // Add post loop after the given loop. Node *insert_post_loop(IdealLoopTree* loop, Node_List& old_new, CountedLoopNode* main_head, CountedLoopEndNode* main_end, Node* incr, Node* limit, CountedLoopNode*& post_head); // Add a vector post loop between a vector main loop and the current post loop void insert_vector_post_loop(IdealLoopTree *loop, Node_List &old_new); // If Node n lives in the back_ctrl block, we clone a private version of n // in preheader_ctrl block and return that, otherwise return n. Node *clone_up_backedge_goo( Node *back_ctrl, Node *preheader_ctrl, Node *n, VectorSet &visited, Node_Stack &clones ); // Take steps to maximally unroll the loop. Peel any odd iterations, then // unroll to do double iterations. The next round of major loop transforms // will repeat till the doubled loop body does all remaining iterations in 1 // pass. void do_maximally_unroll( IdealLoopTree *loop, Node_List &old_new ); // Unroll the loop body one step - make each trip do 2 iterations. void do_unroll( IdealLoopTree *loop, Node_List &old_new, bool adjust_min_trip ); // Return true if exp is a constant times an induction var bool is_scaled_iv(Node* exp, Node* iv, BasicType bt, jlong* p_scale, bool* p_short_scale, int depth = 0); bool is_iv(Node* exp, Node* iv, BasicType bt); // Return true if exp is a scaled induction var plus (or minus) constant bool is_scaled_iv_plus_offset(Node* exp, Node* iv, BasicType bt, jlong* p_scale, Node** p_offset, bool* p_short_scale = nullptr, int depth = 0); bool is_scaled_iv_plus_offset(Node* exp, Node* iv, int* p_scale, Node** p_offset) { jlong long_scale; if (is_scaled_iv_plus_offset(exp, iv, T_INT, &long_scale, p_offset)) { int int_scale = checked_cast(long_scale); if (p_scale != nullptr) { *p_scale = int_scale; } return true; } return false; } // Helper for finding more complex matches to is_scaled_iv_plus_offset. bool is_scaled_iv_plus_extra_offset(Node* exp1, Node* offset2, Node* iv, BasicType bt, jlong* p_scale, Node** p_offset, bool* p_short_scale, int depth); // Create a new if above the uncommon_trap_if_pattern for the predicate to be promoted IfTrueNode* create_new_if_for_predicate(const ParsePredicateSuccessProj* parse_predicate_proj, Node* new_entry, Deoptimization::DeoptReason reason, int opcode, bool rewire_uncommon_proj_phi_inputs = false); private: // Helper functions for create_new_if_for_predicate() void set_ctrl_of_nodes_with_same_ctrl(Node* start_node, ProjNode* old_uncommon_proj, Node* new_uncommon_proj); Unique_Node_List find_nodes_with_same_ctrl(Node* node, const ProjNode* ctrl); Node* clone_nodes_with_same_ctrl(Node* start_node, ProjNode* old_uncommon_proj, Node* new_uncommon_proj); void fix_cloned_data_node_controls(const ProjNode* orig, Node* new_uncommon_proj, const OrigToNewHashtable& orig_to_clone); public: void register_control(Node* n, IdealLoopTree *loop, Node* pred, bool update_body = true); // Replace the control input of 'node' with 'new_control' and set the dom depth to the one of 'new_control'. void replace_control(Node* node, Node* new_control) { _igvn.replace_input_of(node, 0, new_control); set_idom(node, new_control, dom_depth(new_control)); } void replace_loop_entry(LoopNode* loop_head, Node* new_entry) { _igvn.replace_input_of(loop_head, LoopNode::EntryControl, new_entry); set_idom(loop_head, new_entry, dom_depth(new_entry)); } // Construct a range check for a predicate if BoolNode* rc_predicate(Node* ctrl, int scale, Node* offset, Node* init, Node* limit, jint stride, Node* range, bool upper, bool& overflow); // Implementation of the loop predication to promote checks outside the loop bool loop_predication_impl(IdealLoopTree *loop); // Reachability Fence (RF) support. private: void insert_rf(Node* ctrl, Node* referent); void replace_rf(Node* old_node, Node* new_node); void remove_rf(ReachabilityFenceNode* rf); public: bool optimize_reachability_fences(); bool expand_reachability_fences(); private: bool loop_predication_impl_helper(IdealLoopTree* loop, IfProjNode* if_success_proj, ParsePredicateSuccessProj* parse_predicate_proj, CountedLoopNode* cl, ConNode* zero, Invariance& invar, Deoptimization::DeoptReason deopt_reason); bool can_create_loop_predicates(const PredicateBlock* profiled_loop_predicate_block) const; bool loop_predication_should_follow_branches(IdealLoopTree* loop, float& loop_trip_cnt); void loop_predication_follow_branches(Node *c, IdealLoopTree *loop, float loop_trip_cnt, PathFrequency& pf, Node_Stack& stack, VectorSet& seen, Node_List& if_proj_list); IfTrueNode* create_template_assertion_predicate(CountedLoopNode* loop_head, ParsePredicateNode* parse_predicate, IfProjNode* new_control, int scale, Node* offset, Node* range); void eliminate_hoisted_range_check(IfTrueNode* hoisted_check_proj, IfTrueNode* template_assertion_predicate_proj); // Helper function to collect predicate for eliminating the useless ones void eliminate_useless_predicates() const; void eliminate_useless_zero_trip_guard(); void eliminate_useless_multiversion_if(); public: // Change the control input of expensive nodes to allow commoning by // IGVN when it is guaranteed to not result in a more frequent // execution of the expensive node. Return true if progress. bool process_expensive_nodes(); // Check whether node has become unreachable bool is_node_unreachable(Node *n) const { return !has_node(n) || n->is_unreachable(_igvn); } // Eliminate range-checks and other trip-counter vs loop-invariant tests. void do_range_check(IdealLoopTree* loop); // Clone loop with an invariant test (that does not exit) and // insert a clone of the test that selects which version to // execute. void do_unswitching(IdealLoopTree* loop, Node_List& old_new); IfNode* find_unswitch_candidate(const IdealLoopTree* loop) const; private: static bool has_control_dependencies_from_predicates(LoopNode* head); static void revert_to_normal_loop(const LoopNode* loop_head); void hoist_invariant_check_casts(const IdealLoopTree* loop, const Node_List& old_new, const UnswitchedLoopSelector& unswitched_loop_selector); void add_unswitched_loop_version_bodies_to_igvn(IdealLoopTree* loop, const Node_List& old_new); static void increment_unswitch_counts(LoopNode* original_head, LoopNode* new_head); void remove_unswitch_candidate_from_loops(const Node_List& old_new, const UnswitchedLoopSelector& unswitched_loop_selector); #ifndef PRODUCT static void trace_loop_unswitching_count(IdealLoopTree* loop, LoopNode* original_head); static void trace_loop_unswitching_impossible(const LoopNode* original_head); static void trace_loop_unswitching_result(const UnswitchedLoopSelector& unswitched_loop_selector, const LoopNode* original_head, const LoopNode* new_head); static void trace_loop_multiversioning_result(const LoopSelector& loop_selector, const LoopNode* original_head, const LoopNode* new_head); #endif public: // Range Check Elimination uses this function! // Constrain the main loop iterations so the affine function: // low_limit <= scale_con * I + offset < upper_limit // always holds true. That is, either increase the number of iterations in // the pre-loop or the post-loop until the condition holds true in the main // loop. Scale_con, offset and limit are all loop invariant. void add_constraint(jlong stride_con, jlong scale_con, Node* offset, Node* low_limit, Node* upper_limit, Node* pre_ctrl, Node** pre_limit, Node** main_limit); // Helper function for add_constraint(). Node* adjust_limit(bool reduce, Node* scale, Node* offset, Node* rc_limit, Node* old_limit, Node* pre_ctrl, bool round); // Partially peel loop up through last_peel node. bool partial_peel( IdealLoopTree *loop, Node_List &old_new ); bool duplicate_loop_backedge(IdealLoopTree *loop, Node_List &old_new); // AutoVectorize the loop: replace scalar ops with vector ops. enum AutoVectorizeStatus { Impossible, // This loop has the wrong shape to even try vectorization. Success, // We just successfully vectorized the loop. TriedAndFailed, // We tried to vectorize, but failed. }; AutoVectorizeStatus auto_vectorize(IdealLoopTree* lpt, VSharedData &vshared); void maybe_multiversion_for_auto_vectorization_runtime_checks(IdealLoopTree* lpt, Node_List& old_new); void do_multiversioning(IdealLoopTree* lpt, Node_List& old_new); IfTrueNode* create_new_if_for_multiversion(IfTrueNode* multiversioning_fast_proj); bool try_resume_optimizations_for_delayed_slow_loop(IdealLoopTree* lpt); // Create a scheduled list of nodes control dependent on ctrl set. void scheduled_nodelist( IdealLoopTree *loop, VectorSet& ctrl, Node_List &sched ); // Has a use in the vector set bool has_use_in_set( Node* n, VectorSet& vset ); // Has use internal to the vector set (ie. not in a phi at the loop head) bool has_use_internal_to_set( Node* n, VectorSet& vset, IdealLoopTree *loop ); // clone "n" for uses that are outside of loop int clone_for_use_outside_loop( IdealLoopTree *loop, Node* n, Node_List& worklist ); // clone "n" for special uses that are in the not_peeled region void clone_for_special_use_inside_loop( IdealLoopTree *loop, Node* n, VectorSet& not_peel, Node_List& sink_list, Node_List& worklist ); // Insert phi(lp_entry_val, back_edge_val) at use->in(idx) for loop lp if phi does not already exist void insert_phi_for_loop( Node* use, uint idx, Node* lp_entry_val, Node* back_edge_val, LoopNode* lp ); #ifdef ASSERT // Validate the loop partition sets: peel and not_peel bool is_valid_loop_partition( IdealLoopTree *loop, VectorSet& peel, Node_List& peel_list, VectorSet& not_peel ); // Ensure that uses outside of loop are of the right form bool is_valid_clone_loop_form( IdealLoopTree *loop, Node_List& peel_list, uint orig_exit_idx, uint clone_exit_idx); bool is_valid_clone_loop_exit_use( IdealLoopTree *loop, Node* use, uint exit_idx); #endif // Returns nonzero constant stride if-node is a possible iv test (otherwise returns zero.) int stride_of_possible_iv( Node* iff ); bool is_possible_iv_test( Node* iff ) { return stride_of_possible_iv(iff) != 0; } // Return the (unique) control output node that's in the loop (if it exists.) Node* stay_in_loop( Node* n, IdealLoopTree *loop); // Insert a signed compare loop exit cloned from an unsigned compare. IfNode* insert_cmpi_loop_exit(IfNode* if_cmpu, IdealLoopTree *loop); void remove_cmpi_loop_exit(IfNode* if_cmp, IdealLoopTree *loop); // Utility to register node "n" with PhaseIdealLoop void register_node(Node* n, IdealLoopTree* loop, Node* pred, uint ddepth); // Utility to create an if-projection ProjNode* proj_clone(ProjNode* p, IfNode* iff); // Force the iff control output to be the live_proj Node* short_circuit_if(IfNode* iff, ProjNode* live_proj); // Insert a region before an if projection RegionNode* insert_region_before_proj(ProjNode* proj); // Insert a new if before an if projection ProjNode* insert_if_before_proj(Node* left, bool Signed, BoolTest::mask relop, Node* right, ProjNode* proj); // Passed in a Phi merging (recursively) some nearly equivalent Bool/Cmps. // "Nearly" because all Nodes have been cloned from the original in the loop, // but the fall-in edges to the Cmp are different. Clone bool/Cmp pairs // through the Phi recursively, and return a Bool. Node* clone_iff(PhiNode* phi); CmpNode* clone_bool(PhiNode* phi); // Rework addressing expressions to get the most loop-invariant stuff // moved out. We'd like to do all associative operators, but it's especially // important (common) to do address expressions. Node* remix_address_expressions(Node* n); Node* remix_address_expressions_add_left_shift(Node* n, IdealLoopTree* n_loop, Node* n_ctrl, BasicType bt); // Convert add to muladd to generate MuladdS2I under certain criteria Node * convert_add_to_muladd(Node * n); // Attempt to use a conditional move instead of a phi/branch Node *conditional_move( Node *n ); // Check for aggressive application of 'split-if' optimization, // using basic block level info. void split_if_with_blocks ( VectorSet &visited, Node_Stack &nstack); Node *split_if_with_blocks_pre ( Node *n ); void split_if_with_blocks_post( Node *n ); Node *has_local_phi_input( Node *n ); // Mark an IfNode as being dominated by a prior test, // without actually altering the CFG (and hence IDOM info). void dominated_by(IfProjNode* prevdom, IfNode* iff, bool flip = false, bool prev_dom_not_imply_this = false); void rewire_safe_outputs_to_dominator(Node* source, Node* dominator, bool dominator_not_imply_source); // Split Node 'n' through merge point RegionNode* split_thru_region(Node* n, RegionNode* region); // Split Node 'n' through merge point if there is enough win. Node *split_thru_phi( Node *n, Node *region, int policy ); // Found an If getting its condition-code input from a Phi in the // same block. Split thru the Region. void do_split_if(Node *iff, RegionNode** new_false_region = nullptr, RegionNode** new_true_region = nullptr); private: // Class to keep track of wins in split_thru_phi. class SplitThruPhiWins { private: // Region containing the phi we are splitting through. const Node* _region; // Sum of all wins regardless of where they happen. This applies to Loops phis as well as non-loop phis. int _total_wins; // For Loops, wins have different impact depending on if they happen on loop entry or on the backedge. // Number of wins on a loop entry edge if the split is through a loop head, // otherwise 0. Entry edge wins only pay dividends once on loop entry. int _loop_entry_wins; // Number of wins on a loop back-edge, which pay dividends on every iteration. int _loop_back_wins; public: SplitThruPhiWins(const Node* region) : _region(region), _total_wins(0), _loop_entry_wins(0), _loop_back_wins(0) {}; void reset() {_total_wins = 0; _loop_entry_wins = 0; _loop_back_wins = 0;} void add_win(int ctrl_index) { if (_region->is_Loop() && ctrl_index == LoopNode::EntryControl) { _loop_entry_wins++; } else if (_region->is_Loop() && ctrl_index == LoopNode::LoopBackControl) { _loop_back_wins++; } _total_wins++; } // Is this split profitable with respect to the policy? bool profitable(int policy) const { assert(_region->is_Loop() || (_loop_entry_wins == 0 && _loop_back_wins == 0), "wins on loop edges without a loop"); assert(!_region->is_Loop() || _total_wins == _loop_entry_wins + _loop_back_wins, "missed some win"); // In general this means that the split has to have more wins than specified // in the policy. However, for loops we need to take into account where the // wins happen. We need to be careful when splitting, because splitting nodes // related to the iv through the phi can sufficiently rearrange the loop // structure to prevent RCE and thus vectorization. Thus, we only deem splitting // profitable if the win of a split is not on the entry edge, as such wins // only pay off once and have a high chance of messing up the loop structure. return (_loop_entry_wins == 0 && _total_wins > policy) || // If there are wins on the entry edge but the backadge also has sufficient wins, // there is sufficient profitability to spilt regardless of the risk of messing // up the loop structure. _loop_back_wins > policy || // If the policy is less than 0, a split is always profitable, i.e. we always // split. This is needed when we split a node and then must also split a // dependant node, i.e. spliting a Bool node after splitting a Cmp node. policy < 0; } }; void split_thru_phi_yank_old_nodes(Node* n, Node* region); public: // Conversion of fill/copy patterns into intrinsic versions bool do_intrinsify_fill(); bool intrinsify_fill(IdealLoopTree* lpt); bool match_fill_loop(IdealLoopTree* lpt, Node*& store, Node*& store_value, Node*& shift, Node*& offset); private: // Helper functions Node *spinup( Node *iff, Node *new_false, Node *new_true, Node *region, Node *phi, small_cache *cache ); Node *find_use_block( Node *use, Node *def, Node *old_false, Node *new_false, Node *old_true, Node *new_true ); void handle_use( Node *use, Node *def, small_cache *cache, Node *region_dom, Node *new_false, Node *new_true, Node *old_false, Node *old_true ); bool split_up( Node *n, Node *blk1, Node *blk2 ); Node* place_outside_loop(Node* useblock, IdealLoopTree* loop) const; Node* try_move_store_before_loop(Node* n, Node *n_ctrl); void try_move_store_after_loop(Node* n); bool identical_backtoback_ifs(Node *n); bool can_split_if(Node *n_ctrl); bool cannot_split_division(const Node* n, const Node* region) const; static bool is_divisor_loop_phi(const Node* divisor, const Node* loop); bool loop_phi_backedge_type_contains_zero(const Node* phi_divisor, const Type* zero) const; // Determine if a method is too big for a/another round of split-if, based on // a magic (approximate) ratio derived from the equally magic constant 35000, // previously used for this purpose (but without relating to the node limit). bool must_throttle_split_if() { uint threshold = C->max_node_limit() * 2 / 5; return C->live_nodes() > threshold; } // A simplistic node request tracking mechanism, where // = UINT_MAX Request not valid or made final. // < UINT_MAX Nodes currently requested (estimate). uint _nodes_required; enum { REQUIRE_MIN = 70 }; uint nodes_required() const { return _nodes_required; } // Given the _currently_ available number of nodes, check whether there is // "room" for an additional request or not, considering the already required // number of nodes. Return TRUE if the new request is exceeding the node // budget limit, otherwise return FALSE. Note that this interpretation will // act pessimistic on additional requests when new nodes have already been // generated since the 'begin'. This behaviour fits with the intention that // node estimates/requests should be made upfront. bool exceeding_node_budget(uint required = 0) { assert(C->live_nodes() < C->max_node_limit(), "sanity"); uint available = C->max_node_limit() - C->live_nodes(); return available < required + _nodes_required + REQUIRE_MIN; } uint require_nodes(uint require, uint minreq = REQUIRE_MIN) { precond(require > 0); _nodes_required += MAX2(require, minreq); return _nodes_required; } bool may_require_nodes(uint require, uint minreq = REQUIRE_MIN) { return !exceeding_node_budget(require) && require_nodes(require, minreq) > 0; } uint require_nodes_begin() { assert(_nodes_required == UINT_MAX, "Bad state (begin)."); _nodes_required = 0; return C->live_nodes(); } // When a node request is final, optionally check that the requested number // of nodes was reasonably correct with respect to the number of new nodes // introduced since the last 'begin'. Always check that we have not exceeded // the maximum node limit. void require_nodes_final(uint live_at_begin, bool check_estimate) { assert(_nodes_required < UINT_MAX, "Bad state (final)."); #ifdef ASSERT if (check_estimate) { // Check that the node budget request was not off by too much (x2). // Should this be the case we _surely_ need to improve the estimates // used in our budget calculations. if (C->live_nodes() - live_at_begin > 2 * _nodes_required) { log_info(compilation)("Bad node estimate: actual = %d >> request = %d", C->live_nodes() - live_at_begin, _nodes_required); } } #endif // Assert that we have stayed within the node budget limit. assert(C->live_nodes() < C->max_node_limit(), "Exceeding node budget limit: %d + %d > %d (request = %d)", C->live_nodes() - live_at_begin, live_at_begin, C->max_node_limit(), _nodes_required); _nodes_required = UINT_MAX; } private: bool _created_loop_node; DEBUG_ONLY(void dump_idoms(Node* early, Node* wrong_lca);) NOT_PRODUCT(void dump_idoms_in_reverse(const Node* n, const Node_List& idom_list) const;) public: void set_created_loop_node() { _created_loop_node = true; } bool created_loop_node() { return _created_loop_node; } void register_new_node(Node* n, Node* blk); void register_new_node_with_ctrl_of(Node* new_node, Node* ctrl_of) { register_new_node(new_node, get_ctrl(ctrl_of)); } Node* clone_and_register(Node* n, Node* ctrl) { n = n->clone(); register_new_node(n, ctrl); return n; } #ifdef ASSERT void dump_bad_graph(const char* msg, Node* n, Node* early, Node* LCA); #endif #ifndef PRODUCT void dump() const; void dump_idom(Node* n) const { dump_idom(n, 1000); } // For debugging void dump_idom(Node* n, uint count) const; void get_idoms(Node* n, uint count, Unique_Node_List& idoms) const; void dump(IdealLoopTree* loop, uint rpo_idx, Node_List &rpo_list) const; IdealLoopTree* get_loop_idx(Node* n) const { // Dead nodes have no loop, so return the top level loop instead return _loop_or_ctrl[n->_idx] ? (IdealLoopTree*)_loop_or_ctrl[n->_idx] : _ltree_root; } // Print some stats static void print_statistics(); static int _loop_invokes; // Count of PhaseIdealLoop invokes static int _loop_work; // Sum of PhaseIdealLoop x _unique static volatile int _long_loop_candidates; static volatile int _long_loop_nests; #endif #ifdef ASSERT void verify() const; bool verify_idom_and_nodes(Node* root, const PhaseIdealLoop* phase_verify) const; bool verify_idom(Node* n, const PhaseIdealLoop* phase_verify) const; bool verify_loop_ctrl(Node* n, const PhaseIdealLoop* phase_verify) const; #endif void rpo(Node* start, Node_Stack &stk, VectorSet &visited, Node_List &rpo_list) const; void check_counted_loop_shape(IdealLoopTree* loop, Node* head, BasicType bt) NOT_DEBUG_RETURN; LoopNode* create_inner_head(IdealLoopTree* loop, BaseCountedLoopNode* head, IfNode* exit_test); int extract_long_range_checks(const IdealLoopTree* loop, jint stride_con, int iters_limit, PhiNode* phi, Node_List &range_checks); void transform_long_range_checks(int stride_con, const Node_List &range_checks, Node* outer_phi, Node* inner_iters_actual_int, Node* inner_phi, Node* iv_add, LoopNode* inner_head); Node* get_late_ctrl_with_anti_dep(LoadNode* n, Node* early, Node* LCA); bool ctrl_of_use_out_of_loop(const Node* n, Node* n_ctrl, IdealLoopTree* n_loop, Node* ctrl); bool ctrl_of_all_uses_out_of_loop(const Node* n, Node* n_ctrl, IdealLoopTree* n_loop); bool would_sink_below_pre_loop_exit(IdealLoopTree* n_loop, Node* ctrl); Node* compute_early_ctrl(Node* n, Node* n_ctrl); void try_sink_out_of_loop(Node* n); Node* clamp(Node* R, Node* L, Node* H); bool safe_for_if_replacement(const Node* dom) const; void push_pinned_nodes_thru_region(IfNode* dom_if, Node* region); bool try_merge_identical_ifs(Node* n); void clone_loop_body(const Node_List& body, Node_List &old_new, CloneMap* cm); void fix_body_edges(const Node_List &body, IdealLoopTree* loop, const Node_List &old_new, int dd, IdealLoopTree* parent, bool partial); void fix_ctrl_uses(const Node_List& body, const IdealLoopTree* loop, Node_List &old_new, CloneLoopMode mode, Node* side_by_side_idom, CloneMap* cm, Node_List &worklist); void fix_data_uses(Node_List& body, IdealLoopTree* loop, CloneLoopMode mode, IdealLoopTree* outer_loop, uint new_counter, Node_List& old_new, Node_List& worklist, Node_List*& split_if_set, Node_List*& split_bool_set, Node_List*& split_cex_set); void finish_clone_loop(Node_List* split_if_set, Node_List* split_bool_set, Node_List* split_cex_set); bool at_relevant_ctrl(Node* n, const Node* blk1, const Node* blk2); bool clone_cmp_loadklass_down(Node* n, const Node* blk1, const Node* blk2); void clone_loadklass_nodes_at_cmp_index(const Node* n, Node* cmp, int i); bool clone_cmp_down(Node* n, const Node* blk1, const Node* blk2); void clone_template_assertion_expression_down(Node* node); Node* similar_subtype_check(const Node* x, Node* r_in); void update_addp_chain_base(Node* x, Node* old_base, Node* new_base); bool can_move_to_inner_loop(Node* n, LoopNode* n_loop, Node* x); void pin_nodes_dependent_on(Node* ctrl, bool old_iff_is_rangecheck); Node* ensure_node_and_inputs_are_above_pre_end(CountedLoopEndNode* pre_end, Node* node); Node* new_assertion_predicate_opaque_init(Node* entry_control, Node* init, Node* int_zero); bool try_make_short_running_loop(IdealLoopTree* loop, jint stride_con, const Node_List& range_checks, const uint iters_limit); ConINode* intcon(jint i); ConLNode* longcon(jlong i); ConNode* makecon(const Type* t); ConNode* integercon(jlong l, BasicType bt); ConNode* zerocon(BasicType bt); }; class CountedLoopConverter { friend class PhaseIdealLoop; // Match increment with optional truncation class TruncatedIncrement { bool _is_valid; BasicType _bt; Node* _incr; Node* _outer_trunc; Node* _inner_trunc; const TypeInteger* _trunc_type; public: TruncatedIncrement(BasicType bt) : _is_valid(false), _bt(bt), _incr(nullptr), _outer_trunc(nullptr), _inner_trunc(nullptr), _trunc_type(nullptr) {} void build(Node* expr); bool is_valid() const { return _is_valid; } Node* incr() const { return _incr; } // Optional truncation for: CHAR: (i+1)&0x7fff, BYTE: ((i+1)<<8)>>8, or SHORT: ((i+1)<<16)>>16 Node* outer_trunc() const { return _outer_trunc; } // the outermost truncating node (either the & or the final >>) Node* inner_trunc() const { return _inner_trunc; } // the inner truncating node, if applicable (the << in a <> pair) const TypeInteger* trunc_type() const { return _trunc_type; } }; class LoopStructure { bool _is_valid; const Node* _head; const IdealLoopTree* _loop; PhaseIdealLoop* _phase; BasicType _iv_bt; Node* _back_control; PhaseIdealLoop::LoopExitTest _exit_test; PhaseIdealLoop::LoopIVIncr _iv_incr; TruncatedIncrement _truncated_increment; PhaseIdealLoop::LoopIVStride _stride; PhiNode* _phi; SafePointNode* _safepoint; public: LoopStructure(const Node* head, const IdealLoopTree* loop, PhaseIdealLoop* phase, const BasicType iv_bt) : _is_valid(false), _head(head), _loop(loop), _phase(phase), _iv_bt(iv_bt), _back_control(_phase->loop_exit_control(_loop)), _exit_test(_back_control, _loop, _phase), _iv_incr(_head, _loop), _truncated_increment(_iv_bt), _stride(PhaseIdealLoop::LoopIVStride(_iv_bt)), _phi(nullptr), _safepoint(nullptr) {} void build(); jlong final_limit_correction() const; // compute adjusted loop limit correction bool is_infinite_loop() const; bool is_valid() const { return _is_valid; } Node* back_control() const { return _back_control; } PhaseIdealLoop::LoopExitTest& exit_test() { return _exit_test; } PhaseIdealLoop::LoopIVIncr& iv_incr() { return _iv_incr; } TruncatedIncrement& truncated_increment() { return _truncated_increment; } PhaseIdealLoop::LoopIVStride& stride() { return _stride; } PhiNode* phi() const { return _phi; } SafePointNode* sfpt() const { return _safepoint; } jlong stride_con() const { return _stride.compute_non_zero_stride_con(_exit_test.mask(), _iv_bt); } Node* limit() const { return _exit_test.limit(); } }; PhaseIdealLoop* const _phase; Node* const _head; IdealLoopTree* const _loop; const BasicType _iv_bt; LoopStructure _structure; bool _should_insert_stride_overflow_limit_check = false; bool _should_insert_init_trip_limit_check = false; DEBUG_ONLY(bool _checked_for_counted_loop = false;) // stats for PhaseIdealLoop::print_statistics() static volatile int _long_loop_counted_loops; // Return a type based on condition control flow const TypeInt* filtered_type(Node* n, Node* n_ctrl); const TypeInt* filtered_type(Node* n) { return filtered_type(n, nullptr); } // Helpers for filtered type const TypeInt* filtered_type_from_dominators(Node* val, Node* val_ctrl); void insert_loop_limit_check_predicate(const ParsePredicateSuccessProj* loop_limit_check_parse_proj, Node* bol) const; void insert_stride_overflow_limit_check() const; void insert_init_trip_limit_check() const; bool has_dominating_loop_limit_check(Node* init_trip, Node* limit, jlong stride_con, BasicType iv_bt, Node* loop_entry) const; bool is_iv_overflowing(const TypeInteger* init_t, jlong stride_con, Node* phi_increment, BoolTest::mask mask) const; bool has_truncation_wrap(const TruncatedIncrement& truncation, Node* phi, jlong stride_con); SafePointNode* find_safepoint(Node* iftrue); bool is_safepoint_invalid(SafePointNode* sfpt) const; public: CountedLoopConverter(PhaseIdealLoop* phase, Node* head, IdealLoopTree* loop, const BasicType iv_bt) : _phase(phase), _head(head), _loop(loop), _iv_bt(iv_bt), _structure(LoopStructure(_head, _loop, _phase, _iv_bt)) { assert(phase != nullptr, "must be"); // Fail early if mandatory parameters are null. assert(head != nullptr, "must be"); assert(loop != nullptr, "must be"); assert(iv_bt == T_INT || iv_bt == T_LONG, "either int or long loops"); } bool is_counted_loop(); IdealLoopTree* convert(); DEBUG_ONLY(bool should_stress_long_counted_loop();) DEBUG_ONLY(bool stress_long_counted_loop();) enum StrideOverflowState { Overflow = -1, NoOverflow = 0, RequireLimitCheck = 1 }; static StrideOverflowState check_stride_overflow(jlong final_correction, const TypeInteger* limit_t, BasicType bt); }; class AutoNodeBudget : public StackObj { public: enum budget_check_t { BUDGET_CHECK, NO_BUDGET_CHECK }; AutoNodeBudget(PhaseIdealLoop* phase, budget_check_t chk = BUDGET_CHECK) : _phase(phase), _check_at_final(chk == BUDGET_CHECK), _nodes_at_begin(0) { precond(_phase != nullptr); _nodes_at_begin = _phase->require_nodes_begin(); } ~AutoNodeBudget() { #ifndef PRODUCT if (TraceLoopOpts) { uint request = _phase->nodes_required(); uint delta = _phase->C->live_nodes() - _nodes_at_begin; if (request < delta) { tty->print_cr("Exceeding node budget: %d < %d", request, delta); } else { uint const REQUIRE_MIN = PhaseIdealLoop::REQUIRE_MIN; // Identify the worst estimates as "poor" ones. if (request > REQUIRE_MIN && delta > 0) { if ((delta > REQUIRE_MIN && request > 3 * delta) || (delta <= REQUIRE_MIN && request > 10 * delta)) { tty->print_cr("Poor node estimate: %d >> %d", request, delta); } } } } #endif // PRODUCT _phase->require_nodes_final(_nodes_at_begin, _check_at_final); } private: PhaseIdealLoop* _phase; bool _check_at_final; uint _nodes_at_begin; }; inline Node* IdealLoopTree::tail() { // Handle lazy update of _tail field. if (_tail->in(0) == nullptr) { _tail = _phase->get_ctrl(_tail); } return _tail; } inline Node* IdealLoopTree::head() { // Handle lazy update of _head field. if (_head->in(0) == nullptr) { _head = _phase->get_ctrl(_head); } return _head; } // Iterate over the loop tree using a preorder, left-to-right traversal. // // Example that visits all counted loops from within PhaseIdealLoop // // for (LoopTreeIterator iter(_ltree_root); !iter.done(); iter.next()) { // IdealLoopTree* lpt = iter.current(); // if (!lpt->is_counted()) continue; // ... class LoopTreeIterator : public StackObj { private: IdealLoopTree* _root; IdealLoopTree* _curnt; public: LoopTreeIterator(IdealLoopTree* root) : _root(root), _curnt(root) {} bool done() { return _curnt == nullptr; } // Finished iterating? void next(); // Advance to next loop tree IdealLoopTree* current() { return _curnt; } // Return current value of iterator. }; // Compute probability of reaching some CFG node from a fixed // dominating CFG node class PathFrequency { private: Node* _dom; // frequencies are computed relative to this node Node_Stack _stack; GrowableArray _freqs_stack; // keep track of intermediate result at regions GrowableArray _freqs; // cache frequencies PhaseIdealLoop* _phase; float check_and_truncate_frequency(float f) { assert(f >= 0, "Incorrect frequency"); // We do not perform an exact (f <= 1) check // this would be error prone with rounding of floats. // Performing a check like (f <= 1+eps) would be of benefit, // however, it is not evident how to determine such an eps, // given that an arbitrary number of add/mul operations // are performed on these frequencies. return (f > 1) ? 1 : f; } public: PathFrequency(Node* dom, PhaseIdealLoop* phase) : _dom(dom), _stack(0), _phase(phase) { } float to(Node* n); }; // Class to clone a data node graph by taking a list of data nodes. This is done in 2 steps: // 1. Clone the data nodes // 2. Fix the cloned data inputs pointing to the old nodes to the cloned inputs by using an old->new mapping. class DataNodeGraph : public StackObj { PhaseIdealLoop* const _phase; const Unique_Node_List& _data_nodes; OrigToNewHashtable _orig_to_new; public: DataNodeGraph(const Unique_Node_List& data_nodes, PhaseIdealLoop* phase) : _phase(phase), _data_nodes(data_nodes), // Use 107 as best guess which is the first resize value in ResizeableHashTable::large_table_sizes. _orig_to_new(107, MaxNodeLimit) { #ifdef ASSERT for (uint i = 0; i < data_nodes.size(); i++) { assert(!data_nodes[i]->is_CFG(), "only data nodes"); } #endif } NONCOPYABLE(DataNodeGraph); private: void clone(Node* node, Node* new_ctrl); void clone_data_nodes(Node* new_ctrl); void clone_data_nodes_and_transform_opaque_loop_nodes(const TransformStrategyForOpaqueLoopNodes& transform_strategy, Node* new_ctrl); void rewire_clones_to_cloned_inputs(); void transform_opaque_node(const TransformStrategyForOpaqueLoopNodes& transform_strategy, Node* node); public: // Clone the provided data node collection and rewire the clones in such a way to create an identical graph copy. // Set 'new_ctrl' as ctrl for the cloned nodes. const OrigToNewHashtable& clone(Node* new_ctrl) { assert(_orig_to_new.number_of_entries() == 0, "should not call this method twice in a row"); clone_data_nodes(new_ctrl); rewire_clones_to_cloned_inputs(); return _orig_to_new; } // Create a copy of the data nodes provided to the constructor by doing the following: // Clone all non-OpaqueLoop* nodes and rewire them to create an identical subgraph copy. For the OpaqueLoop* nodes, // apply the provided transformation strategy and include the transformed node into the subgraph copy to get a complete // "cloned-and-transformed" graph copy. For all newly cloned nodes (which could also be new OpaqueLoop* nodes), set // `new_ctrl` as ctrl. const OrigToNewHashtable& clone_with_opaque_loop_transform_strategy( const TransformStrategyForOpaqueLoopNodes& transform_strategy, Node* new_ctrl) { clone_data_nodes_and_transform_opaque_loop_nodes(transform_strategy, new_ctrl); rewire_clones_to_cloned_inputs(); return _orig_to_new; } }; #endif // SHARE_OPTO_LOOPNODE_HPP