8198515: Extract SoftReferencePolicy code out of CollectorPolicy
Reviewed-by: pliden, sjohanss
/*
* Copyright (c) 2000, 2018, 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/cardTableModRefBS.inline.hpp"
#include "gc/shared/collectedHeap.hpp"
#include "gc/shared/genCollectedHeap.hpp"
#include "gc/shared/space.inline.hpp"
#include "logging/log.hpp"
#include "memory/virtualspace.hpp"
#include "oops/oop.inline.hpp"
#include "runtime/thread.hpp"
#include "services/memTracker.hpp"
#include "utilities/align.hpp"
#include "utilities/macros.hpp"
// This kind of "BarrierSet" allows a "CollectedHeap" to detect and
// enumerate ref fields that have been modified (since the last
// enumeration.)
size_t CardTableModRefBS::compute_byte_map_size()
{
assert(_guard_index == cards_required(_whole_heap.word_size()) - 1,
"uninitialized, check declaration order");
assert(_page_size != 0, "uninitialized, check declaration order");
const size_t granularity = os::vm_allocation_granularity();
return align_up(_guard_index + 1, MAX2(_page_size, granularity));
}
CardTableModRefBS::CardTableModRefBS(
MemRegion whole_heap,
const BarrierSet::FakeRtti& fake_rtti) :
ModRefBarrierSet(fake_rtti.add_tag(BarrierSet::CardTableModRef)),
_whole_heap(whole_heap),
_guard_index(0),
_guard_region(),
_last_valid_index(0),
_page_size(os::vm_page_size()),
_byte_map_size(0),
_covered(NULL),
_committed(NULL),
_cur_covered_regions(0),
_byte_map(NULL),
byte_map_base(NULL),
_defer_initial_card_mark(false)
{
assert((uintptr_t(_whole_heap.start()) & (card_size - 1)) == 0, "heap must start at card boundary");
assert((uintptr_t(_whole_heap.end()) & (card_size - 1)) == 0, "heap must end at card boundary");
assert(card_size <= 512, "card_size must be less than 512"); // why?
_covered = new MemRegion[_max_covered_regions];
if (_covered == NULL) {
vm_exit_during_initialization("Could not allocate card table covered region set.");
}
}
void CardTableModRefBS::initialize() {
initialize_deferred_card_mark_barriers();
_guard_index = cards_required(_whole_heap.word_size()) - 1;
_last_valid_index = _guard_index - 1;
_byte_map_size = compute_byte_map_size();
HeapWord* low_bound = _whole_heap.start();
HeapWord* high_bound = _whole_heap.end();
_cur_covered_regions = 0;
_committed = new MemRegion[_max_covered_regions];
if (_committed == NULL) {
vm_exit_during_initialization("Could not allocate card table committed region set.");
}
const size_t rs_align = _page_size == (size_t) os::vm_page_size() ? 0 :
MAX2(_page_size, (size_t) os::vm_allocation_granularity());
ReservedSpace heap_rs(_byte_map_size, rs_align, false);
MemTracker::record_virtual_memory_type((address)heap_rs.base(), mtGC);
os::trace_page_sizes("Card Table", _guard_index + 1, _guard_index + 1,
_page_size, heap_rs.base(), heap_rs.size());
if (!heap_rs.is_reserved()) {
vm_exit_during_initialization("Could not reserve enough space for the "
"card marking array");
}
// The assembler store_check code will do an unsigned shift of the oop,
// then add it to byte_map_base, i.e.
//
// _byte_map = byte_map_base + (uintptr_t(low_bound) >> card_shift)
_byte_map = (jbyte*) heap_rs.base();
byte_map_base = _byte_map - (uintptr_t(low_bound) >> card_shift);
assert(byte_for(low_bound) == &_byte_map[0], "Checking start of map");
assert(byte_for(high_bound-1) <= &_byte_map[_last_valid_index], "Checking end of map");
jbyte* guard_card = &_byte_map[_guard_index];
uintptr_t guard_page = align_down((uintptr_t)guard_card, _page_size);
_guard_region = MemRegion((HeapWord*)guard_page, _page_size);
os::commit_memory_or_exit((char*)guard_page, _page_size, _page_size,
!ExecMem, "card table last card");
*guard_card = last_card;
log_trace(gc, barrier)("CardTableModRefBS::CardTableModRefBS: ");
log_trace(gc, barrier)(" &_byte_map[0]: " INTPTR_FORMAT " &_byte_map[_last_valid_index]: " INTPTR_FORMAT,
p2i(&_byte_map[0]), p2i(&_byte_map[_last_valid_index]));
log_trace(gc, barrier)(" byte_map_base: " INTPTR_FORMAT, p2i(byte_map_base));
}
CardTableModRefBS::~CardTableModRefBS() {
if (_covered) {
delete[] _covered;
_covered = NULL;
}
if (_committed) {
delete[] _committed;
_committed = NULL;
}
}
int CardTableModRefBS::find_covering_region_by_base(HeapWord* base) {
int i;
for (i = 0; i < _cur_covered_regions; i++) {
if (_covered[i].start() == base) return i;
if (_covered[i].start() > base) break;
}
// If we didn't find it, create a new one.
assert(_cur_covered_regions < _max_covered_regions,
"too many covered regions");
// Move the ones above up, to maintain sorted order.
for (int j = _cur_covered_regions; j > i; j--) {
_covered[j] = _covered[j-1];
_committed[j] = _committed[j-1];
}
int res = i;
_cur_covered_regions++;
_covered[res].set_start(base);
_covered[res].set_word_size(0);
jbyte* ct_start = byte_for(base);
uintptr_t ct_start_aligned = align_down((uintptr_t)ct_start, _page_size);
_committed[res].set_start((HeapWord*)ct_start_aligned);
_committed[res].set_word_size(0);
return res;
}
int CardTableModRefBS::find_covering_region_containing(HeapWord* addr) {
for (int i = 0; i < _cur_covered_regions; i++) {
if (_covered[i].contains(addr)) {
return i;
}
}
assert(0, "address outside of heap?");
return -1;
}
HeapWord* CardTableModRefBS::largest_prev_committed_end(int ind) const {
HeapWord* max_end = NULL;
for (int j = 0; j < ind; j++) {
HeapWord* this_end = _committed[j].end();
if (this_end > max_end) max_end = this_end;
}
return max_end;
}
MemRegion CardTableModRefBS::committed_unique_to_self(int self,
MemRegion mr) const {
MemRegion result = mr;
for (int r = 0; r < _cur_covered_regions; r += 1) {
if (r != self) {
result = result.minus(_committed[r]);
}
}
// Never include the guard page.
result = result.minus(_guard_region);
return result;
}
void CardTableModRefBS::resize_covered_region(MemRegion new_region) {
// We don't change the start of a region, only the end.
assert(_whole_heap.contains(new_region),
"attempt to cover area not in reserved area");
debug_only(verify_guard();)
// collided is true if the expansion would push into another committed region
debug_only(bool collided = false;)
int const ind = find_covering_region_by_base(new_region.start());
MemRegion const old_region = _covered[ind];
assert(old_region.start() == new_region.start(), "just checking");
if (new_region.word_size() != old_region.word_size()) {
// Commit new or uncommit old pages, if necessary.
MemRegion cur_committed = _committed[ind];
// Extend the end of this _committed region
// to cover the end of any lower _committed regions.
// This forms overlapping regions, but never interior regions.
HeapWord* const max_prev_end = largest_prev_committed_end(ind);
if (max_prev_end > cur_committed.end()) {
cur_committed.set_end(max_prev_end);
}
// Align the end up to a page size (starts are already aligned).
jbyte* const new_end = byte_after(new_region.last());
HeapWord* new_end_aligned = (HeapWord*) align_up(new_end, _page_size);
assert((void*)new_end_aligned >= (void*) new_end, "align up, but less");
// Check the other regions (excludes "ind") to ensure that
// the new_end_aligned does not intrude onto the committed
// space of another region.
int ri = 0;
for (ri = ind + 1; ri < _cur_covered_regions; ri++) {
if (new_end_aligned > _committed[ri].start()) {
assert(new_end_aligned <= _committed[ri].end(),
"An earlier committed region can't cover a later committed region");
// Any region containing the new end
// should start at or beyond the region found (ind)
// for the new end (committed regions are not expected to
// be proper subsets of other committed regions).
assert(_committed[ri].start() >= _committed[ind].start(),
"New end of committed region is inconsistent");
new_end_aligned = _committed[ri].start();
// new_end_aligned can be equal to the start of its
// committed region (i.e., of "ind") if a second
// region following "ind" also start at the same location
// as "ind".
assert(new_end_aligned >= _committed[ind].start(),
"New end of committed region is before start");
debug_only(collided = true;)
// Should only collide with 1 region
break;
}
}
#ifdef ASSERT
for (++ri; ri < _cur_covered_regions; ri++) {
assert(!_committed[ri].contains(new_end_aligned),
"New end of committed region is in a second committed region");
}
#endif
// The guard page is always committed and should not be committed over.
// "guarded" is used for assertion checking below and recalls the fact
// that the would-be end of the new committed region would have
// penetrated the guard page.
HeapWord* new_end_for_commit = new_end_aligned;
DEBUG_ONLY(bool guarded = false;)
if (new_end_for_commit > _guard_region.start()) {
new_end_for_commit = _guard_region.start();
DEBUG_ONLY(guarded = true;)
}
if (new_end_for_commit > cur_committed.end()) {
// Must commit new pages.
MemRegion const new_committed =
MemRegion(cur_committed.end(), new_end_for_commit);
assert(!new_committed.is_empty(), "Region should not be empty here");
os::commit_memory_or_exit((char*)new_committed.start(),
new_committed.byte_size(), _page_size,
!ExecMem, "card table expansion");
// Use new_end_aligned (as opposed to new_end_for_commit) because
// the cur_committed region may include the guard region.
} else if (new_end_aligned < cur_committed.end()) {
// Must uncommit pages.
MemRegion const uncommit_region =
committed_unique_to_self(ind, MemRegion(new_end_aligned,
cur_committed.end()));
if (!uncommit_region.is_empty()) {
// It is not safe to uncommit cards if the boundary between
// the generations is moving. A shrink can uncommit cards
// owned by generation A but being used by generation B.
if (!UseAdaptiveGCBoundary) {
if (!os::uncommit_memory((char*)uncommit_region.start(),
uncommit_region.byte_size())) {
assert(false, "Card table contraction failed");
// The call failed so don't change the end of the
// committed region. This is better than taking the
// VM down.
new_end_aligned = _committed[ind].end();
}
} else {
new_end_aligned = _committed[ind].end();
}
}
}
// In any case, we can reset the end of the current committed entry.
_committed[ind].set_end(new_end_aligned);
#ifdef ASSERT
// Check that the last card in the new region is committed according
// to the tables.
bool covered = false;
for (int cr = 0; cr < _cur_covered_regions; cr++) {
if (_committed[cr].contains(new_end - 1)) {
covered = true;
break;
}
}
assert(covered, "Card for end of new region not committed");
#endif
// The default of 0 is not necessarily clean cards.
jbyte* entry;
if (old_region.last() < _whole_heap.start()) {
entry = byte_for(_whole_heap.start());
} else {
entry = byte_after(old_region.last());
}
assert(index_for(new_region.last()) < _guard_index,
"The guard card will be overwritten");
// This line commented out cleans the newly expanded region and
// not the aligned up expanded region.
// jbyte* const end = byte_after(new_region.last());
jbyte* const end = (jbyte*) new_end_for_commit;
assert((end >= byte_after(new_region.last())) || collided || guarded,
"Expect to be beyond new region unless impacting another region");
// do nothing if we resized downward.
#ifdef ASSERT
for (int ri = 0; ri < _cur_covered_regions; ri++) {
if (ri != ind) {
// The end of the new committed region should not
// be in any existing region unless it matches
// the start of the next region.
assert(!_committed[ri].contains(end) ||
(_committed[ri].start() == (HeapWord*) end),
"Overlapping committed regions");
}
}
#endif
if (entry < end) {
memset(entry, clean_card, pointer_delta(end, entry, sizeof(jbyte)));
}
}
// In any case, the covered size changes.
_covered[ind].set_word_size(new_region.word_size());
log_trace(gc, barrier)("CardTableModRefBS::resize_covered_region: ");
log_trace(gc, barrier)(" _covered[%d].start(): " INTPTR_FORMAT " _covered[%d].last(): " INTPTR_FORMAT,
ind, p2i(_covered[ind].start()), ind, p2i(_covered[ind].last()));
log_trace(gc, barrier)(" _committed[%d].start(): " INTPTR_FORMAT " _committed[%d].last(): " INTPTR_FORMAT,
ind, p2i(_committed[ind].start()), ind, p2i(_committed[ind].last()));
log_trace(gc, barrier)(" byte_for(start): " INTPTR_FORMAT " byte_for(last): " INTPTR_FORMAT,
p2i(byte_for(_covered[ind].start())), p2i(byte_for(_covered[ind].last())));
log_trace(gc, barrier)(" addr_for(start): " INTPTR_FORMAT " addr_for(last): " INTPTR_FORMAT,
p2i(addr_for((jbyte*) _committed[ind].start())), p2i(addr_for((jbyte*) _committed[ind].last())));
// Touch the last card of the covered region to show that it
// is committed (or SEGV).
debug_only((void) (*byte_for(_covered[ind].last()));)
debug_only(verify_guard();)
}
// Note that these versions are precise! The scanning code has to handle the
// fact that the write barrier may be either precise or imprecise.
void CardTableModRefBS::dirty_MemRegion(MemRegion mr) {
assert(align_down(mr.start(), HeapWordSize) == mr.start(), "Unaligned start");
assert(align_up (mr.end(), HeapWordSize) == mr.end(), "Unaligned end" );
jbyte* cur = byte_for(mr.start());
jbyte* last = byte_after(mr.last());
while (cur < last) {
*cur = dirty_card;
cur++;
}
}
void CardTableModRefBS::invalidate(MemRegion mr) {
assert(align_down(mr.start(), HeapWordSize) == mr.start(), "Unaligned start");
assert(align_up (mr.end(), HeapWordSize) == mr.end(), "Unaligned end" );
for (int i = 0; i < _cur_covered_regions; i++) {
MemRegion mri = mr.intersection(_covered[i]);
if (!mri.is_empty()) dirty_MemRegion(mri);
}
}
void CardTableModRefBS::clear_MemRegion(MemRegion mr) {
// Be conservative: only clean cards entirely contained within the
// region.
jbyte* cur;
if (mr.start() == _whole_heap.start()) {
cur = byte_for(mr.start());
} else {
assert(mr.start() > _whole_heap.start(), "mr is not covered.");
cur = byte_after(mr.start() - 1);
}
jbyte* last = byte_after(mr.last());
memset(cur, clean_card, pointer_delta(last, cur, sizeof(jbyte)));
}
void CardTableModRefBS::clear(MemRegion mr) {
for (int i = 0; i < _cur_covered_regions; i++) {
MemRegion mri = mr.intersection(_covered[i]);
if (!mri.is_empty()) clear_MemRegion(mri);
}
}
void CardTableModRefBS::dirty(MemRegion mr) {
jbyte* first = byte_for(mr.start());
jbyte* last = byte_after(mr.last());
memset(first, dirty_card, last-first);
}
// Unlike several other card table methods, dirty_card_iterate()
// iterates over dirty cards ranges in increasing address order.
void CardTableModRefBS::dirty_card_iterate(MemRegion mr,
MemRegionClosure* cl) {
for (int i = 0; i < _cur_covered_regions; i++) {
MemRegion mri = mr.intersection(_covered[i]);
if (!mri.is_empty()) {
jbyte *cur_entry, *next_entry, *limit;
for (cur_entry = byte_for(mri.start()), limit = byte_for(mri.last());
cur_entry <= limit;
cur_entry = next_entry) {
next_entry = cur_entry + 1;
if (*cur_entry == dirty_card) {
size_t dirty_cards;
// Accumulate maximal dirty card range, starting at cur_entry
for (dirty_cards = 1;
next_entry <= limit && *next_entry == dirty_card;
dirty_cards++, next_entry++);
MemRegion cur_cards(addr_for(cur_entry),
dirty_cards*card_size_in_words);
cl->do_MemRegion(cur_cards);
}
}
}
}
}
MemRegion CardTableModRefBS::dirty_card_range_after_reset(MemRegion mr,
bool reset,
int reset_val) {
for (int i = 0; i < _cur_covered_regions; i++) {
MemRegion mri = mr.intersection(_covered[i]);
if (!mri.is_empty()) {
jbyte* cur_entry, *next_entry, *limit;
for (cur_entry = byte_for(mri.start()), limit = byte_for(mri.last());
cur_entry <= limit;
cur_entry = next_entry) {
next_entry = cur_entry + 1;
if (*cur_entry == dirty_card) {
size_t dirty_cards;
// Accumulate maximal dirty card range, starting at cur_entry
for (dirty_cards = 1;
next_entry <= limit && *next_entry == dirty_card;
dirty_cards++, next_entry++);
MemRegion cur_cards(addr_for(cur_entry),
dirty_cards*card_size_in_words);
if (reset) {
for (size_t i = 0; i < dirty_cards; i++) {
cur_entry[i] = reset_val;
}
}
return cur_cards;
}
}
}
}
return MemRegion(mr.end(), mr.end());
}
uintx CardTableModRefBS::ct_max_alignment_constraint() {
return card_size * os::vm_page_size();
}
void CardTableModRefBS::verify_guard() {
// For product build verification
guarantee(_byte_map[_guard_index] == last_card,
"card table guard has been modified");
}
void CardTableModRefBS::verify() {
verify_guard();
}
#ifndef PRODUCT
void CardTableModRefBS::verify_region(MemRegion mr,
jbyte val, bool val_equals) {
jbyte* start = byte_for(mr.start());
jbyte* end = byte_for(mr.last());
bool failures = false;
for (jbyte* curr = start; curr <= end; ++curr) {
jbyte curr_val = *curr;
bool failed = (val_equals) ? (curr_val != val) : (curr_val == val);
if (failed) {
if (!failures) {
log_error(gc, verify)("== CT verification failed: [" INTPTR_FORMAT "," INTPTR_FORMAT "]", p2i(start), p2i(end));
log_error(gc, verify)("== %sexpecting value: %d", (val_equals) ? "" : "not ", val);
failures = true;
}
log_error(gc, verify)("== card " PTR_FORMAT " [" PTR_FORMAT "," PTR_FORMAT "], val: %d",
p2i(curr), p2i(addr_for(curr)),
p2i((HeapWord*) (((size_t) addr_for(curr)) + card_size)),
(int) curr_val);
}
}
guarantee(!failures, "there should not have been any failures");
}
void CardTableModRefBS::verify_not_dirty_region(MemRegion mr) {
verify_region(mr, dirty_card, false /* val_equals */);
}
void CardTableModRefBS::verify_dirty_region(MemRegion mr) {
verify_region(mr, dirty_card, true /* val_equals */);
}
#endif
void CardTableModRefBS::print_on(outputStream* st) const {
st->print_cr("Card table byte_map: [" INTPTR_FORMAT "," INTPTR_FORMAT "] byte_map_base: " INTPTR_FORMAT,
p2i(_byte_map), p2i(_byte_map + _byte_map_size), p2i(byte_map_base));
}
// Helper for ReduceInitialCardMarks. For performance,
// compiled code may elide card-marks for initializing stores
// to a newly allocated object along the fast-path. We
// compensate for such elided card-marks as follows:
// (a) Generational, non-concurrent collectors, such as
// GenCollectedHeap(ParNew,DefNew,Tenured) and
// ParallelScavengeHeap(ParallelGC, ParallelOldGC)
// need the card-mark if and only if the region is
// in the old gen, and do not care if the card-mark
// succeeds or precedes the initializing stores themselves,
// so long as the card-mark is completed before the next
// scavenge. For all these cases, we can do a card mark
// at the point at which we do a slow path allocation
// in the old gen, i.e. in this call.
// (b) GenCollectedHeap(ConcurrentMarkSweepGeneration) requires
// in addition that the card-mark for an old gen allocated
// object strictly follow any associated initializing stores.
// In these cases, the memRegion remembered below is
// used to card-mark the entire region either just before the next
// slow-path allocation by this thread or just before the next scavenge or
// CMS-associated safepoint, whichever of these events happens first.
// (The implicit assumption is that the object has been fully
// initialized by this point, a fact that we assert when doing the
// card-mark.)
// (c) G1CollectedHeap(G1) uses two kinds of write barriers. When a
// G1 concurrent marking is in progress an SATB (pre-write-)barrier
// is used to remember the pre-value of any store. Initializing
// stores will not need this barrier, so we need not worry about
// compensating for the missing pre-barrier here. Turning now
// to the post-barrier, we note that G1 needs a RS update barrier
// which simply enqueues a (sequence of) dirty cards which may
// optionally be refined by the concurrent update threads. Note
// that this barrier need only be applied to a non-young write,
// but, like in CMS, because of the presence of concurrent refinement
// (much like CMS' precleaning), must strictly follow the oop-store.
// Thus, using the same protocol for maintaining the intended
// invariants turns out, serendepitously, to be the same for both
// G1 and CMS.
//
// For any future collector, this code should be reexamined with
// that specific collector in mind, and the documentation above suitably
// extended and updated.
void CardTableModRefBS::on_slowpath_allocation_exit(JavaThread* thread, oop new_obj) {
if (!ReduceInitialCardMarks) {
return;
}
// If a previous card-mark was deferred, flush it now.
flush_deferred_card_mark_barrier(thread);
if (new_obj->is_typeArray() || is_in_young(new_obj)) {
// Arrays of non-references don't need a post-barrier.
// The deferred_card_mark region should be empty
// following the flush above.
assert(thread->deferred_card_mark().is_empty(), "Error");
} else {
MemRegion mr((HeapWord*)new_obj, new_obj->size());
assert(!mr.is_empty(), "Error");
if (_defer_initial_card_mark) {
// Defer the card mark
thread->set_deferred_card_mark(mr);
} else {
// Do the card mark
write_region(mr);
}
}
}
void CardTableModRefBS::initialize_deferred_card_mark_barriers() {
// Used for ReduceInitialCardMarks (when COMPILER2 or JVMCI is used);
// otherwise remains unused.
#if defined(COMPILER2) || INCLUDE_JVMCI
_defer_initial_card_mark = is_server_compilation_mode_vm() && ReduceInitialCardMarks && can_elide_tlab_store_barriers()
&& (DeferInitialCardMark || card_mark_must_follow_store());
#else
assert(_defer_initial_card_mark == false, "Who would set it?");
#endif
}
void CardTableModRefBS::flush_deferred_card_mark_barrier(JavaThread* thread) {
#if defined(COMPILER2) || INCLUDE_JVMCI
MemRegion deferred = thread->deferred_card_mark();
if (!deferred.is_empty()) {
assert(_defer_initial_card_mark, "Otherwise should be empty");
{
// Verify that the storage points to a parsable object in heap
DEBUG_ONLY(oop old_obj = oop(deferred.start());)
assert(!is_in_young(old_obj),
"Else should have been filtered in on_slowpath_allocation_exit()");
assert(oopDesc::is_oop(old_obj, true), "Not an oop");
assert(deferred.word_size() == (size_t)(old_obj->size()),
"Mismatch: multiple objects?");
}
write_region(deferred);
// "Clear" the deferred_card_mark field
thread->set_deferred_card_mark(MemRegion());
}
assert(thread->deferred_card_mark().is_empty(), "invariant");
#else
assert(!_defer_initial_card_mark, "Should be false");
assert(thread->deferred_card_mark().is_empty(), "Should be empty");
#endif
}
void CardTableModRefBS::on_thread_detach(JavaThread* thread) {
// The deferred store barriers must all have been flushed to the
// card-table (or other remembered set structure) before GC starts
// processing the card-table (or other remembered set).
flush_deferred_card_mark_barrier(thread);
}