/*

 * Copyright (c) 2001, 2017, 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.

 *

 */

#include "precompiled.hpp"

#include "gc/shared/cardTableRS.hpp"

#include "gc/shared/genCollectedHeap.hpp"

#include "gc/shared/generation.hpp"

#include "gc/shared/space.inline.hpp"

#include "memory/allocation.inline.hpp"

#include "oops/oop.inline.hpp"

#include "runtime/atomic.hpp"

#include "runtime/java.hpp"

#include "runtime/os.hpp"

#include "utilities/macros.hpp"

class HasAccumulatedModifiedOopsClosure : public CLDClosure {

  bool _found;

 public:

  HasAccumulatedModifiedOopsClosure() : _found(false) {}

  void do_cld(ClassLoaderData* cld) {

    if (_found) {

      return;

    }

    if (cld->has_accumulated_modified_oops()) {

      _found = true;

    }

  }

  bool found() {

    return _found;

  }

};

bool CLDRemSet::mod_union_is_clear() {

  HasAccumulatedModifiedOopsClosure closure;

  ClassLoaderDataGraph::cld_do(&closure);

  return !closure.found();

}

class ClearCLDModUnionClosure : public CLDClosure {

 public:

  void do_cld(ClassLoaderData* cld) {

    if (cld->has_accumulated_modified_oops()) {

      cld->clear_accumulated_modified_oops();

    }

  }

};

void CLDRemSet::clear_mod_union() {

  ClearCLDModUnionClosure closure;

  ClassLoaderDataGraph::cld_do(&closure);

}

CardTableRS::CardTableRS(MemRegion whole_heap) :

  _bs(NULL),

  _cur_youngergen_card_val(youngergenP1_card)

{

  _ct_bs = new CardTableModRefBSForCTRS(whole_heap);

  _ct_bs->initialize();

  set_bs(_ct_bs);

  // max_gens is really GenCollectedHeap::heap()->gen_policy()->number_of_generations()

  // (which is always 2, young & old), but GenCollectedHeap has not been initialized yet.

  uint max_gens = 2;

  _last_cur_val_in_gen = NEW_C_HEAP_ARRAY3(jbyte, max_gens + 1,

                         mtGC, CURRENT_PC, AllocFailStrategy::RETURN_NULL);

  if (_last_cur_val_in_gen == NULL) {

    vm_exit_during_initialization("Could not create last_cur_val_in_gen array.");

  }

  for (uint i = 0; i < max_gens + 1; i++) {

    _last_cur_val_in_gen[i] = clean_card_val();

  }

  _ct_bs->set_CTRS(this);

}

CardTableRS::~CardTableRS() {

  if (_ct_bs) {

    delete _ct_bs;

    _ct_bs = NULL;

  }

  if (_last_cur_val_in_gen) {

    FREE_C_HEAP_ARRAY(jbyte, _last_cur_val_in_gen);

  }

}

void CardTableRS::resize_covered_region(MemRegion new_region) {

  _ct_bs->resize_covered_region(new_region);

}

jbyte CardTableRS::find_unused_youngergenP_card_value() {

  for (jbyte v = youngergenP1_card;

       v < cur_youngergen_and_prev_nonclean_card;

       v++) {

    bool seen = false;

    for (int g = 0; g < _regions_to_iterate; g++) {

      if (_last_cur_val_in_gen[g] == v) {

        seen = true;

        break;

      }

    }

    if (!seen) {

      return v;

    }

  }

  ShouldNotReachHere();

  return 0;

}

void CardTableRS::prepare_for_younger_refs_iterate(bool parallel) {

  // Parallel or sequential, we must always set the prev to equal the

  // last one written.

  if (parallel) {

    // Find a parallel value to be used next.

    jbyte next_val = find_unused_youngergenP_card_value();

    set_cur_youngergen_card_val(next_val);

  } else {

    // In an sequential traversal we will always write youngergen, so that

    // the inline barrier is  correct.

    set_cur_youngergen_card_val(youngergen_card);

  }

}

void CardTableRS::younger_refs_iterate(Generation* g,

                                       OopsInGenClosure* blk,

                                       uint n_threads) {

  // The indexing in this array is slightly odd. We want to access

  // the old generation record here, which is at index 2.

  _last_cur_val_in_gen[2] = cur_youngergen_card_val();

  g->younger_refs_iterate(blk, n_threads);

}

inline bool ClearNoncleanCardWrapper::clear_card(jbyte* entry) {

  if (_is_par) {

    return clear_card_parallel(entry);

  } else {

    return clear_card_serial(entry);

  }

}

inline bool ClearNoncleanCardWrapper::clear_card_parallel(jbyte* entry) {

  while (true) {

    // In the parallel case, we may have to do this several times.

    jbyte entry_val = *entry;

    assert(entry_val != CardTableRS::clean_card_val(),

           "We shouldn't be looking at clean cards, and this should "

           "be the only place they get cleaned.");

    if (CardTableRS::card_is_dirty_wrt_gen_iter(entry_val)

        || _ct->is_prev_youngergen_card_val(entry_val)) {

      jbyte res =

        Atomic::cmpxchg(CardTableRS::clean_card_val(), entry, entry_val);

      if (res == entry_val) {

        break;

      } else {

        assert(res == CardTableRS::cur_youngergen_and_prev_nonclean_card,

               "The CAS above should only fail if another thread did "

               "a GC write barrier.");

      }

    } else if (entry_val ==

               CardTableRS::cur_youngergen_and_prev_nonclean_card) {

      // Parallelism shouldn't matter in this case.  Only the thread

      // assigned to scan the card should change this value.

      *entry = _ct->cur_youngergen_card_val();

      break;

    } else {

      assert(entry_val == _ct->cur_youngergen_card_val(),

             "Should be the only possibility.");

      // In this case, the card was clean before, and become

      // cur_youngergen only because of processing of a promoted object.

      // We don't have to look at the card.

      return false;

    }

  }

  return true;

}

inline bool ClearNoncleanCardWrapper::clear_card_serial(jbyte* entry) {

  jbyte entry_val = *entry;

  assert(entry_val != CardTableRS::clean_card_val(),

         "We shouldn't be looking at clean cards, and this should "

         "be the only place they get cleaned.");

  assert(entry_val != CardTableRS::cur_youngergen_and_prev_nonclean_card,

         "This should be possible in the sequential case.");

  *entry = CardTableRS::clean_card_val();

  return true;

}

ClearNoncleanCardWrapper::ClearNoncleanCardWrapper(

  DirtyCardToOopClosure* dirty_card_closure, CardTableRS* ct, bool is_par) :

    _dirty_card_closure(dirty_card_closure), _ct(ct), _is_par(is_par) {

}

bool ClearNoncleanCardWrapper::is_word_aligned(jbyte* entry) {

  return (((intptr_t)entry) & (BytesPerWord-1)) == 0;

}

// The regions are visited in *decreasing* address order.

// This order aids with imprecise card marking, where a dirty

// card may cause scanning, and summarization marking, of objects

// that extend onto subsequent cards.

void ClearNoncleanCardWrapper::do_MemRegion(MemRegion mr) {

  assert(mr.word_size() > 0, "Error");

  assert(_ct->is_aligned(mr.start()), "mr.start() should be card aligned");

  // mr.end() may not necessarily be card aligned.

  jbyte* cur_entry = _ct->byte_for(mr.last());

  const jbyte* limit = _ct->byte_for(mr.start());

  HeapWord* end_of_non_clean = mr.end();

  HeapWord* start_of_non_clean = end_of_non_clean;

  while (cur_entry >= limit) {

    HeapWord* cur_hw = _ct->addr_for(cur_entry);

    if ((*cur_entry != CardTableRS::clean_card_val()) && clear_card(cur_entry)) {

      // Continue the dirty range by opening the

      // dirty window one card to the left.

      start_of_non_clean = cur_hw;

    } else {

      // We hit a "clean" card; process any non-empty

      // "dirty" range accumulated so far.

      if (start_of_non_clean < end_of_non_clean) {

        const MemRegion mrd(start_of_non_clean, end_of_non_clean);

        _dirty_card_closure->do_MemRegion(mrd);

      }

      // fast forward through potential continuous whole-word range of clean cards beginning at a word-boundary

      if (is_word_aligned(cur_entry)) {

        jbyte* cur_row = cur_entry - BytesPerWord;

        while (cur_row >= limit && *((intptr_t*)cur_row) ==  CardTableRS::clean_card_row()) {

          cur_row -= BytesPerWord;

        }

        cur_entry = cur_row + BytesPerWord;

        cur_hw = _ct->addr_for(cur_entry);

      }

      // Reset the dirty window, while continuing to look

      // for the next dirty card that will start a

      // new dirty window.

      end_of_non_clean = cur_hw;

      start_of_non_clean = cur_hw;

    }

    // Note that "cur_entry" leads "start_of_non_clean" in

    // its leftward excursion after this point

    // in the loop and, when we hit the left end of "mr",

    // will point off of the left end of the card-table

    // for "mr".

    cur_entry--;

  }

  // If the first card of "mr" was dirty, we will have

  // been left with a dirty window, co-initial with "mr",

  // which we now process.

  if (start_of_non_clean < end_of_non_clean) {

    const MemRegion mrd(start_of_non_clean, end_of_non_clean);

    _dirty_card_closure->do_MemRegion(mrd);

  }

}

// clean (by dirty->clean before) ==> cur_younger_gen

// dirty                          ==> cur_youngergen_and_prev_nonclean_card

// precleaned                     ==> cur_youngergen_and_prev_nonclean_card

// prev-younger-gen               ==> cur_youngergen_and_prev_nonclean_card

// cur-younger-gen                ==> cur_younger_gen

// cur_youngergen_and_prev_nonclean_card ==> no change.

void CardTableRS::write_ref_field_gc_par(void* field, oop new_val) {

  volatile jbyte* entry = _ct_bs->byte_for(field);

  do {

    jbyte entry_val = *entry;

    // We put this first because it's probably the most common case.

    if (entry_val == clean_card_val()) {

      // No threat of contention with cleaning threads.

      *entry = cur_youngergen_card_val();

      return;

    } else if (card_is_dirty_wrt_gen_iter(entry_val)

               || is_prev_youngergen_card_val(entry_val)) {

      // Mark it as both cur and prev youngergen; card cleaning thread will

      // eventually remove the previous stuff.

      jbyte new_val = cur_youngergen_and_prev_nonclean_card;

      jbyte res = Atomic::cmpxchg(new_val, entry, entry_val);

      // Did the CAS succeed?

      if (res == entry_val) return;

      // Otherwise, retry, to see the new value.

      continue;

    } else {

      assert(entry_val == cur_youngergen_and_prev_nonclean_card

             || entry_val == cur_youngergen_card_val(),

             "should be only possibilities.");

      return;

    }

  } while (true);

}

void CardTableRS::younger_refs_in_space_iterate(Space* sp,

                                                OopsInGenClosure* cl,

                                                uint n_threads) {

  const MemRegion urasm = sp->used_region_at_save_marks();

#ifdef ASSERT

  // Convert the assertion check to a warning if we are running

  // CMS+ParNew until related bug is fixed.

  MemRegion ur    = sp->used_region();

  assert(ur.contains(urasm) || (UseConcMarkSweepGC),

         "Did you forget to call save_marks()? "

         "[" PTR_FORMAT ", " PTR_FORMAT ") is not contained in "

         "[" PTR_FORMAT ", " PTR_FORMAT ")",

         p2i(urasm.start()), p2i(urasm.end()), p2i(ur.start()), p2i(ur.end()));

  // In the case of CMS+ParNew, issue a warning

  if (!ur.contains(urasm)) {

    assert(UseConcMarkSweepGC, "Tautology: see assert above");

    log_warning(gc)("CMS+ParNew: Did you forget to call save_marks()? "

                    "[" PTR_FORMAT ", " PTR_FORMAT ") is not contained in "

                    "[" PTR_FORMAT ", " PTR_FORMAT ")",

                    p2i(urasm.start()), p2i(urasm.end()), p2i(ur.start()), p2i(ur.end()));

    MemRegion ur2 = sp->used_region();

    MemRegion urasm2 = sp->used_region_at_save_marks();

    if (!ur.equals(ur2)) {

      log_warning(gc)("CMS+ParNew: Flickering used_region()!!");

    }

    if (!urasm.equals(urasm2)) {

      log_warning(gc)("CMS+ParNew: Flickering used_region_at_save_marks()!!");

    }

    ShouldNotReachHere();

  }

#endif

  _ct_bs->non_clean_card_iterate_possibly_parallel(sp, urasm, cl, this, n_threads);

}

void CardTableRS::clear_into_younger(Generation* old_gen) {

  assert(GenCollectedHeap::heap()->is_old_gen(old_gen),

         "Should only be called for the old generation");

  // The card tables for the youngest gen need never be cleared.

  // There's a bit of subtlety in the clear() and invalidate()

  // methods that we exploit here and in invalidate_or_clear()

  // below to avoid missing cards at the fringes. If clear() or

  // invalidate() are changed in the future, this code should

  // be revisited. 20040107.ysr

  clear(old_gen->prev_used_region());

}

void CardTableRS::invalidate_or_clear(Generation* old_gen) {

  assert(GenCollectedHeap::heap()->is_old_gen(old_gen),

         "Should only be called for the old generation");

  // Invalidate the cards for the currently occupied part of

  // the old generation and clear the cards for the

  // unoccupied part of the generation (if any, making use

  // of that generation's prev_used_region to determine that

  // region). No need to do anything for the youngest

  // generation. Also see note#20040107.ysr above.

  MemRegion used_mr = old_gen->used_region();

  MemRegion to_be_cleared_mr = old_gen->prev_used_region().minus(used_mr);

  if (!to_be_cleared_mr.is_empty()) {

    clear(to_be_cleared_mr);

  }

  invalidate(used_mr);

}

class VerifyCleanCardClosure: public OopClosure {

private:

  HeapWord* _boundary;

  HeapWord* _begin;

  HeapWord* _end;

protected:

  template <class T> void do_oop_work(T* p) {

    HeapWord* jp = (HeapWord*)p;

    assert(jp >= _begin && jp < _end,

           "Error: jp " PTR_FORMAT " should be within "

           "[_begin, _end) = [" PTR_FORMAT "," PTR_FORMAT ")",

           p2i(jp), p2i(_begin), p2i(_end));

    oop obj = oopDesc::load_decode_heap_oop(p);

    guarantee(obj == NULL || (HeapWord*)obj >= _boundary,

              "pointer " PTR_FORMAT " at " PTR_FORMAT " on "

              "clean card crosses boundary" PTR_FORMAT,

              p2i(obj), p2i(jp), p2i(_boundary));

  }

public:

  VerifyCleanCardClosure(HeapWord* b, HeapWord* begin, HeapWord* end) :

    _boundary(b), _begin(begin), _end(end) {

    assert(b <= begin,

           "Error: boundary " PTR_FORMAT " should be at or below begin " PTR_FORMAT,

           p2i(b), p2i(begin));

    assert(begin <= end,

           "Error: begin " PTR_FORMAT " should be strictly below end " PTR_FORMAT,

           p2i(begin), p2i(end));

  }

  virtual void do_oop(oop* p)       { VerifyCleanCardClosure::do_oop_work(p); }

  virtual void do_oop(narrowOop* p) { VerifyCleanCardClosure::do_oop_work(p); }

};

class VerifyCTSpaceClosure: public SpaceClosure {

private:

  CardTableRS* _ct;

  HeapWord* _boundary;

public:

  VerifyCTSpaceClosure(CardTableRS* ct, HeapWord* boundary) :

    _ct(ct), _boundary(boundary) {}

  virtual void do_space(Space* s) { _ct->verify_space(s, _boundary); }

};

class VerifyCTGenClosure: public GenCollectedHeap::GenClosure {

  CardTableRS* _ct;

public:

  VerifyCTGenClosure(CardTableRS* ct) : _ct(ct) {}

  void do_generation(Generation* gen) {

    // Skip the youngest generation.

    if (GenCollectedHeap::heap()->is_young_gen(gen)) {

      return;

    }

    // Normally, we're interested in pointers to younger generations.

    VerifyCTSpaceClosure blk(_ct, gen->reserved().start());

    gen->space_iterate(&blk, true);

  }

};

void CardTableRS::verify_space(Space* s, HeapWord* gen_boundary) {

  // We don't need to do young-gen spaces.

  if (s->end() <= gen_boundary) return;

  MemRegion used = s->used_region();

  jbyte* cur_entry = byte_for(used.start());

  jbyte* limit = byte_after(used.last());

  while (cur_entry < limit) {

    if (*cur_entry == clean_card_val()) {

      jbyte* first_dirty = cur_entry+1;

      while (first_dirty < limit &&

             *first_dirty == clean_card_val()) {

        first_dirty++;

      }

      // If the first object is a regular object, and it has a

      // young-to-old field, that would mark the previous card.

      HeapWord* boundary = addr_for(cur_entry);

      HeapWord* end = (first_dirty >= limit) ? used.end() : addr_for(first_dirty);

      HeapWord* boundary_block = s->block_start(boundary);

      HeapWord* begin = boundary;             // Until proven otherwise.

      HeapWord* start_block = boundary_block; // Until proven otherwise.

      if (boundary_block < boundary) {

        if (s->block_is_obj(boundary_block) && s->obj_is_alive(boundary_block)) {

          oop boundary_obj = oop(boundary_block);

          if (!boundary_obj->is_objArray() &&

              !boundary_obj->is_typeArray()) {

            guarantee(cur_entry > byte_for(used.start()),

                      "else boundary would be boundary_block");

            if (*byte_for(boundary_block) != clean_card_val()) {

              begin = boundary_block + s->block_size(boundary_block);

              start_block = begin;

            }

          }

        }

      }

      // Now traverse objects until end.

      if (begin < end) {

        MemRegion mr(begin, end);

        VerifyCleanCardClosure verify_blk(gen_boundary, begin, end);

        for (HeapWord* cur = start_block; cur < end; cur += s->block_size(cur)) {

          if (s->block_is_obj(cur) && s->obj_is_alive(cur)) {

            oop(cur)->oop_iterate_no_header(&verify_blk, mr);

          }

        }

      }

      cur_entry = first_dirty;

    } else {

      // We'd normally expect that cur_youngergen_and_prev_nonclean_card

      // is a transient value, that cannot be in the card table

      // except during GC, and thus assert that:

      // guarantee(*cur_entry != cur_youngergen_and_prev_nonclean_card,

      //        "Illegal CT value");

      // That however, need not hold, as will become clear in the

      // following...

      // We'd normally expect that if we are in the parallel case,

      // we can't have left a prev value (which would be different

      // from the current value) in the card table, and so we'd like to

      // assert that:

      // guarantee(cur_youngergen_card_val() == youngergen_card

      //           || !is_prev_youngergen_card_val(*cur_entry),

      //           "Illegal CT value");

      // That, however, may not hold occasionally, because of

      // CMS or MSC in the old gen. To wit, consider the