| 1 | /* |
| 2 | * Copyright (c) 2001, 2019, Oracle and/or its affiliates. All rights reserved. |
| 3 | * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER. |
| 4 | * |
| 5 | * This code is free software; you can redistribute it and/or modify it |
| 6 | * under the terms of the GNU General Public License version 2 only, as |
| 7 | * published by the Free Software Foundation. |
| 8 | * |
| 9 | * This code is distributed in the hope that it will be useful, but WITHOUT |
| 10 | * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or |
| 11 | * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License |
| 12 | * version 2 for more details (a copy is included in the LICENSE file that |
| 13 | * accompanied this code). |
| 14 | * |
| 15 | * You should have received a copy of the GNU General Public License version |
| 16 | * 2 along with this work; if not, write to the Free Software Foundation, |
| 17 | * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA. |
| 18 | * |
| 19 | * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA |
| 20 | * or visit www.oracle.com if you need additional information or have any |
| 21 | * questions. |
| 22 | * |
| 23 | */ |
| 24 | |
| 25 | #include "precompiled.hpp" |
| 26 | #include "gc/g1/g1CollectedHeap.inline.hpp" |
| 27 | #include "gc/g1/g1CollectionSetCandidates.hpp" |
| 28 | #include "gc/g1/g1CollectionSetChooser.hpp" |
| 29 | #include "gc/g1/heapRegionRemSet.hpp" |
| 30 | #include "gc/shared/space.inline.hpp" |
| 31 | #include "runtime/atomic.hpp" |
| 32 | #include "utilities/quickSort.hpp" |
| 33 | |
| 34 | // Order regions according to GC efficiency. This will cause regions with a lot |
| 35 | // of live objects and large remembered sets to end up at the end of the array. |
| 36 | // Given that we might skip collecting the last few old regions, if after a few |
| 37 | // mixed GCs the remaining have reclaimable bytes under a certain threshold, the |
| 38 | // hope is that the ones we'll skip are ones with both large remembered sets and |
| 39 | // a lot of live objects, not the ones with just a lot of live objects if we |
| 40 | // ordered according to the amount of reclaimable bytes per region. |
| 41 | static int order_regions(HeapRegion* hr1, HeapRegion* hr2) { |
| 42 | // Make sure that NULL entries are moved to the end. |
| 43 | if (hr1 == NULL) { |
| 44 | if (hr2 == NULL) { |
| 45 | return 0; |
| 46 | } else { |
| 47 | return 1; |
| 48 | } |
| 49 | } else if (hr2 == NULL) { |
| 50 | return -1; |
| 51 | } |
| 52 | |
| 53 | double gc_eff1 = hr1->gc_efficiency(); |
| 54 | double gc_eff2 = hr2->gc_efficiency(); |
| 55 | |
| 56 | if (gc_eff1 > gc_eff2) { |
| 57 | return -1; |
| 58 | } if (gc_eff1 < gc_eff2) { |
| 59 | return 1; |
| 60 | } else { |
| 61 | return 0; |
| 62 | } |
| 63 | } |
| 64 | |
| 65 | // Determine collection set candidates: For all regions determine whether they |
| 66 | // should be a collection set candidates, calculate their efficiency, sort and |
| 67 | // return them as G1CollectionSetCandidates instance. |
| 68 | // Threads calculate the GC efficiency of the regions they get to process, and |
| 69 | // put them into some work area unsorted. At the end the array is sorted and |
| 70 | // copied into the G1CollectionSetCandidates instance; the caller will be the new |
| 71 | // owner of this object. |
| 72 | class G1BuildCandidateRegionsTask : public AbstractGangTask { |
| 73 | |
| 74 | // Work area for building the set of collection set candidates. Contains references |
| 75 | // to heap regions with their GC efficiencies calculated. To reduce contention |
| 76 | // on claiming array elements, worker threads claim parts of this array in chunks; |
| 77 | // Array elements may be NULL as threads might not get enough regions to fill |
| 78 | // up their chunks completely. |
| 79 | // Final sorting will remove them. |
| 80 | class G1BuildCandidateArray : public StackObj { |
| 81 | |
| 82 | uint const _max_size; |
| 83 | uint const _chunk_size; |
| 84 | |
| 85 | HeapRegion** _data; |
| 86 | |
| 87 | uint volatile _cur_claim_idx; |
| 88 | |
| 89 | // Calculates the maximum array size that will be used. |
| 90 | static uint required_array_size(uint num_regions, uint chunk_size, uint num_workers) { |
| 91 | uint const max_waste = num_workers * chunk_size; |
| 92 | // The array should be aligned with respect to chunk_size. |
| 93 | uint const aligned_num_regions = ((num_regions + chunk_size - 1) / chunk_size) * chunk_size; |
| 94 | |
| 95 | return aligned_num_regions + max_waste; |
| 96 | } |
| 97 | |
| 98 | public: |
| 99 | G1BuildCandidateArray(uint max_num_regions, uint chunk_size, uint num_workers) : |
| 100 | _max_size(required_array_size(max_num_regions, chunk_size, num_workers)), |
| 101 | _chunk_size(chunk_size), |
| 102 | _data(NEW_C_HEAP_ARRAY(HeapRegion*, _max_size, mtGC)), |
| 103 | _cur_claim_idx(0) { |
| 104 | for (uint i = 0; i < _max_size; i++) { |
| 105 | _data[i] = NULL; |
| 106 | } |
| 107 | } |
| 108 | |
| 109 | ~G1BuildCandidateArray() { |
| 110 | FREE_C_HEAP_ARRAY(HeapRegion*, _data); |
| 111 | } |
| 112 | |
| 113 | // Claim a new chunk, returning its bounds [from, to[. |
| 114 | void claim_chunk(uint& from, uint& to) { |
| 115 | uint result = Atomic::add(_chunk_size, &_cur_claim_idx); |
| 116 | assert(_max_size > result - 1, |
| 117 | "Array too small, is %u should be %u with chunk size %u." , |
| 118 | _max_size, result, _chunk_size); |
| 119 | from = result - _chunk_size; |
| 120 | to = result; |
| 121 | } |
| 122 | |
| 123 | // Set element in array. |
| 124 | void set(uint idx, HeapRegion* hr) { |
| 125 | assert(idx < _max_size, "Index %u out of bounds %u" , idx, _max_size); |
| 126 | assert(_data[idx] == NULL, "Value must not have been set." ); |
| 127 | _data[idx] = hr; |
| 128 | } |
| 129 | |
| 130 | void sort_and_copy_into(HeapRegion** dest, uint num_regions) { |
| 131 | if (_cur_claim_idx == 0) { |
| 132 | return; |
| 133 | } |
| 134 | for (uint i = _cur_claim_idx; i < _max_size; i++) { |
| 135 | assert(_data[i] == NULL, "must be" ); |
| 136 | } |
| 137 | QuickSort::sort(_data, _cur_claim_idx, order_regions, true); |
| 138 | for (uint i = num_regions; i < _max_size; i++) { |
| 139 | assert(_data[i] == NULL, "must be" ); |
| 140 | } |
| 141 | for (uint i = 0; i < num_regions; i++) { |
| 142 | dest[i] = _data[i]; |
| 143 | } |
| 144 | } |
| 145 | }; |
| 146 | |
| 147 | // Per-region closure. In addition to determining whether a region should be |
| 148 | // added to the candidates, and calculating those regions' gc efficiencies, also |
| 149 | // gather additional statistics. |
| 150 | class G1BuildCandidateRegionsClosure : public HeapRegionClosure { |
| 151 | G1BuildCandidateArray* _array; |
| 152 | |
| 153 | uint _cur_chunk_idx; |
| 154 | uint _cur_chunk_end; |
| 155 | |
| 156 | uint _regions_added; |
| 157 | size_t _reclaimable_bytes_added; |
| 158 | |
| 159 | void add_region(HeapRegion* hr) { |
| 160 | if (_cur_chunk_idx == _cur_chunk_end) { |
| 161 | _array->claim_chunk(_cur_chunk_idx, _cur_chunk_end); |
| 162 | } |
| 163 | assert(_cur_chunk_idx < _cur_chunk_end, "Must be" ); |
| 164 | |
| 165 | hr->calc_gc_efficiency(); |
| 166 | _array->set(_cur_chunk_idx, hr); |
| 167 | |
| 168 | _cur_chunk_idx++; |
| 169 | |
| 170 | _regions_added++; |
| 171 | _reclaimable_bytes_added += hr->reclaimable_bytes(); |
| 172 | } |
| 173 | |
| 174 | bool should_add(HeapRegion* hr) { return G1CollectionSetChooser::should_add(hr); } |
| 175 | |
| 176 | public: |
| 177 | G1BuildCandidateRegionsClosure(G1BuildCandidateArray* array) : |
| 178 | _array(array), |
| 179 | _cur_chunk_idx(0), |
| 180 | _cur_chunk_end(0), |
| 181 | _regions_added(0), |
| 182 | _reclaimable_bytes_added(0) { } |
| 183 | |
| 184 | bool do_heap_region(HeapRegion* r) { |
| 185 | // We will skip any region that's currently used as an old GC |
| 186 | // alloc region (we should not consider those for collection |
| 187 | // before we fill them up). |
| 188 | if (should_add(r) && !G1CollectedHeap::heap()->is_old_gc_alloc_region(r)) { |
| 189 | add_region(r); |
| 190 | } else if (r->is_old()) { |
| 191 | // Keep remembered sets for humongous regions, otherwise clean out remembered |
| 192 | // sets for old regions. |
| 193 | r->rem_set()->clear(true /* only_cardset */); |
| 194 | } else { |
| 195 | assert(r->is_archive() || !r->is_old() || !r->rem_set()->is_tracked(), |
| 196 | "Missed to clear unused remembered set of region %u (%s) that is %s" , |
| 197 | r->hrm_index(), r->get_type_str(), r->rem_set()->get_state_str()); |
| 198 | } |
| 199 | return false; |
| 200 | } |
| 201 | |
| 202 | uint regions_added() const { return _regions_added; } |
| 203 | size_t reclaimable_bytes_added() const { return _reclaimable_bytes_added; } |
| 204 | }; |
| 205 | |
| 206 | G1CollectedHeap* _g1h; |
| 207 | HeapRegionClaimer _hrclaimer; |
| 208 | |
| 209 | uint volatile _num_regions_added; |
| 210 | size_t volatile _reclaimable_bytes_added; |
| 211 | |
| 212 | G1BuildCandidateArray _result; |
| 213 | |
| 214 | void update_totals(uint num_regions, size_t reclaimable_bytes) { |
| 215 | if (num_regions > 0) { |
| 216 | assert(reclaimable_bytes > 0, "invariant" ); |
| 217 | Atomic::add(num_regions, &_num_regions_added); |
| 218 | Atomic::add(reclaimable_bytes, &_reclaimable_bytes_added); |
| 219 | } else { |
| 220 | assert(reclaimable_bytes == 0, "invariant" ); |
| 221 | } |
| 222 | } |
| 223 | |
| 224 | public: |
| 225 | G1BuildCandidateRegionsTask(uint max_num_regions, uint chunk_size, uint num_workers) : |
| 226 | AbstractGangTask("G1 Build Candidate Regions" ), |
| 227 | _g1h(G1CollectedHeap::heap()), |
| 228 | _hrclaimer(num_workers), |
| 229 | _num_regions_added(0), |
| 230 | _reclaimable_bytes_added(0), |
| 231 | _result(max_num_regions, chunk_size, num_workers) { } |
| 232 | |
| 233 | void work(uint worker_id) { |
| 234 | G1BuildCandidateRegionsClosure cl(&_result); |
| 235 | _g1h->heap_region_par_iterate_from_worker_offset(&cl, &_hrclaimer, worker_id); |
| 236 | update_totals(cl.regions_added(), cl.reclaimable_bytes_added()); |
| 237 | } |
| 238 | |
| 239 | G1CollectionSetCandidates* get_sorted_candidates() { |
| 240 | HeapRegion** regions = NEW_C_HEAP_ARRAY(HeapRegion*, _num_regions_added, mtGC); |
| 241 | _result.sort_and_copy_into(regions, _num_regions_added); |
| 242 | return new G1CollectionSetCandidates(regions, |
| 243 | _num_regions_added, |
| 244 | _reclaimable_bytes_added); |
| 245 | } |
| 246 | }; |
| 247 | |
| 248 | uint G1CollectionSetChooser::calculate_work_chunk_size(uint num_workers, uint num_regions) { |
| 249 | assert(num_workers > 0, "Active gc workers should be greater than 0" ); |
| 250 | return MAX2(num_regions / num_workers, 1U); |
| 251 | } |
| 252 | |
| 253 | bool G1CollectionSetChooser::should_add(HeapRegion* hr) { |
| 254 | return !hr->is_young() && |
| 255 | !hr->is_pinned() && |
| 256 | region_occupancy_low_enough_for_evac(hr->live_bytes()) && |
| 257 | hr->rem_set()->is_complete(); |
| 258 | } |
| 259 | |
| 260 | G1CollectionSetCandidates* G1CollectionSetChooser::build(WorkGang* workers, uint max_num_regions) { |
| 261 | uint num_workers = workers->active_workers(); |
| 262 | uint chunk_size = calculate_work_chunk_size(num_workers, max_num_regions); |
| 263 | |
| 264 | G1BuildCandidateRegionsTask cl(max_num_regions, chunk_size, num_workers); |
| 265 | workers->run_task(&cl, num_workers); |
| 266 | |
| 267 | G1CollectionSetCandidates* result = cl.get_sorted_candidates(); |
| 268 | result->verify(); |
| 269 | return result; |
| 270 | } |
| 271 | |