Mercurial > hg > graal-compiler
annotate src/share/vm/gc_implementation/g1/concurrentMark.cpp @ 1974:fd1d227ef1b9
6983204: G1: Nightly test nsk/regression/b4958615 failing with +ExplicitGCInvokesConcurrent
Summary: Enable reference discovery during concurrent marking by setting the reference processor field of the concurrent marking closure. Keep reference objects on the discovered reference lists alive during incremental evacuation pauses until they are processed at the end of concurrent marking.
Reviewed-by: ysr, tonyp
author | johnc |
---|---|
date | Wed, 01 Dec 2010 17:34:02 -0800 |
parents | f95d63e2154a |
children | b03260081e9b |
rev | line source |
---|---|
342 | 1 /* |
1835
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
2 * Copyright (c) 2001, 2010, Oracle and/or its affiliates. All rights reserved. |
342 | 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 * | |
1552
c18cbe5936b8
6941466: Oracle rebranding changes for Hotspot repositories
trims
parents:
1547
diff
changeset
|
19 * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA |
c18cbe5936b8
6941466: Oracle rebranding changes for Hotspot repositories
trims
parents:
1547
diff
changeset
|
20 * or visit www.oracle.com if you need additional information or have any |
c18cbe5936b8
6941466: Oracle rebranding changes for Hotspot repositories
trims
parents:
1547
diff
changeset
|
21 * questions. |
342 | 22 * |
23 */ | |
24 | |
1972 | 25 #include "precompiled.hpp" |
26 #include "classfile/symbolTable.hpp" | |
27 #include "gc_implementation/g1/concurrentMark.hpp" | |
28 #include "gc_implementation/g1/concurrentMarkThread.inline.hpp" | |
29 #include "gc_implementation/g1/g1CollectedHeap.inline.hpp" | |
30 #include "gc_implementation/g1/g1CollectorPolicy.hpp" | |
31 #include "gc_implementation/g1/g1RemSet.hpp" | |
32 #include "gc_implementation/g1/heapRegionRemSet.hpp" | |
33 #include "gc_implementation/g1/heapRegionSeq.inline.hpp" | |
34 #include "memory/genOopClosures.inline.hpp" | |
35 #include "memory/referencePolicy.hpp" | |
36 #include "memory/resourceArea.hpp" | |
37 #include "oops/oop.inline.hpp" | |
38 #include "runtime/handles.inline.hpp" | |
39 #include "runtime/java.hpp" | |
342 | 40 |
41 // | |
42 // CMS Bit Map Wrapper | |
43 | |
44 CMBitMapRO::CMBitMapRO(ReservedSpace rs, int shifter): | |
45 _bm((uintptr_t*)NULL,0), | |
46 _shifter(shifter) { | |
47 _bmStartWord = (HeapWord*)(rs.base()); | |
48 _bmWordSize = rs.size()/HeapWordSize; // rs.size() is in bytes | |
49 ReservedSpace brs(ReservedSpace::allocation_align_size_up( | |
50 (_bmWordSize >> (_shifter + LogBitsPerByte)) + 1)); | |
51 | |
52 guarantee(brs.is_reserved(), "couldn't allocate CMS bit map"); | |
53 // For now we'll just commit all of the bit map up fromt. | |
54 // Later on we'll try to be more parsimonious with swap. | |
55 guarantee(_virtual_space.initialize(brs, brs.size()), | |
56 "couldn't reseve backing store for CMS bit map"); | |
57 assert(_virtual_space.committed_size() == brs.size(), | |
58 "didn't reserve backing store for all of CMS bit map?"); | |
59 _bm.set_map((uintptr_t*)_virtual_space.low()); | |
60 assert(_virtual_space.committed_size() << (_shifter + LogBitsPerByte) >= | |
61 _bmWordSize, "inconsistency in bit map sizing"); | |
62 _bm.set_size(_bmWordSize >> _shifter); | |
63 } | |
64 | |
65 HeapWord* CMBitMapRO::getNextMarkedWordAddress(HeapWord* addr, | |
66 HeapWord* limit) const { | |
67 // First we must round addr *up* to a possible object boundary. | |
68 addr = (HeapWord*)align_size_up((intptr_t)addr, | |
69 HeapWordSize << _shifter); | |
70 size_t addrOffset = heapWordToOffset(addr); | |
71 if (limit == NULL) limit = _bmStartWord + _bmWordSize; | |
72 size_t limitOffset = heapWordToOffset(limit); | |
73 size_t nextOffset = _bm.get_next_one_offset(addrOffset, limitOffset); | |
74 HeapWord* nextAddr = offsetToHeapWord(nextOffset); | |
75 assert(nextAddr >= addr, "get_next_one postcondition"); | |
76 assert(nextAddr == limit || isMarked(nextAddr), | |
77 "get_next_one postcondition"); | |
78 return nextAddr; | |
79 } | |
80 | |
81 HeapWord* CMBitMapRO::getNextUnmarkedWordAddress(HeapWord* addr, | |
82 HeapWord* limit) const { | |
83 size_t addrOffset = heapWordToOffset(addr); | |
84 if (limit == NULL) limit = _bmStartWord + _bmWordSize; | |
85 size_t limitOffset = heapWordToOffset(limit); | |
86 size_t nextOffset = _bm.get_next_zero_offset(addrOffset, limitOffset); | |
87 HeapWord* nextAddr = offsetToHeapWord(nextOffset); | |
88 assert(nextAddr >= addr, "get_next_one postcondition"); | |
89 assert(nextAddr == limit || !isMarked(nextAddr), | |
90 "get_next_one postcondition"); | |
91 return nextAddr; | |
92 } | |
93 | |
94 int CMBitMapRO::heapWordDiffToOffsetDiff(size_t diff) const { | |
95 assert((diff & ((1 << _shifter) - 1)) == 0, "argument check"); | |
96 return (int) (diff >> _shifter); | |
97 } | |
98 | |
99 bool CMBitMapRO::iterate(BitMapClosure* cl, MemRegion mr) { | |
100 HeapWord* left = MAX2(_bmStartWord, mr.start()); | |
101 HeapWord* right = MIN2(_bmStartWord + _bmWordSize, mr.end()); | |
102 if (right > left) { | |
103 // Right-open interval [leftOffset, rightOffset). | |
104 return _bm.iterate(cl, heapWordToOffset(left), heapWordToOffset(right)); | |
105 } else { | |
106 return true; | |
107 } | |
108 } | |
109 | |
110 void CMBitMapRO::mostly_disjoint_range_union(BitMap* from_bitmap, | |
111 size_t from_start_index, | |
112 HeapWord* to_start_word, | |
113 size_t word_num) { | |
114 _bm.mostly_disjoint_range_union(from_bitmap, | |
115 from_start_index, | |
116 heapWordToOffset(to_start_word), | |
117 word_num); | |
118 } | |
119 | |
120 #ifndef PRODUCT | |
121 bool CMBitMapRO::covers(ReservedSpace rs) const { | |
122 // assert(_bm.map() == _virtual_space.low(), "map inconsistency"); | |
645
c3a720eefe82
6816308: Changes to allow builds with latest Windows SDK 6.1 on 64bit Windows 2003
kvn
parents:
619
diff
changeset
|
123 assert(((size_t)_bm.size() * (size_t)(1 << _shifter)) == _bmWordSize, |
342 | 124 "size inconsistency"); |
125 return _bmStartWord == (HeapWord*)(rs.base()) && | |
126 _bmWordSize == rs.size()>>LogHeapWordSize; | |
127 } | |
128 #endif | |
129 | |
130 void CMBitMap::clearAll() { | |
131 _bm.clear(); | |
132 return; | |
133 } | |
134 | |
135 void CMBitMap::markRange(MemRegion mr) { | |
136 mr.intersection(MemRegion(_bmStartWord, _bmWordSize)); | |
137 assert(!mr.is_empty(), "unexpected empty region"); | |
138 assert((offsetToHeapWord(heapWordToOffset(mr.end())) == | |
139 ((HeapWord *) mr.end())), | |
140 "markRange memory region end is not card aligned"); | |
141 // convert address range into offset range | |
142 _bm.at_put_range(heapWordToOffset(mr.start()), | |
143 heapWordToOffset(mr.end()), true); | |
144 } | |
145 | |
146 void CMBitMap::clearRange(MemRegion mr) { | |
147 mr.intersection(MemRegion(_bmStartWord, _bmWordSize)); | |
148 assert(!mr.is_empty(), "unexpected empty region"); | |
149 // convert address range into offset range | |
150 _bm.at_put_range(heapWordToOffset(mr.start()), | |
151 heapWordToOffset(mr.end()), false); | |
152 } | |
153 | |
154 MemRegion CMBitMap::getAndClearMarkedRegion(HeapWord* addr, | |
155 HeapWord* end_addr) { | |
156 HeapWord* start = getNextMarkedWordAddress(addr); | |
157 start = MIN2(start, end_addr); | |
158 HeapWord* end = getNextUnmarkedWordAddress(start); | |
159 end = MIN2(end, end_addr); | |
160 assert(start <= end, "Consistency check"); | |
161 MemRegion mr(start, end); | |
162 if (!mr.is_empty()) { | |
163 clearRange(mr); | |
164 } | |
165 return mr; | |
166 } | |
167 | |
168 CMMarkStack::CMMarkStack(ConcurrentMark* cm) : | |
169 _base(NULL), _cm(cm) | |
170 #ifdef ASSERT | |
171 , _drain_in_progress(false) | |
172 , _drain_in_progress_yields(false) | |
173 #endif | |
174 {} | |
175 | |
176 void CMMarkStack::allocate(size_t size) { | |
177 _base = NEW_C_HEAP_ARRAY(oop, size); | |
178 if (_base == NULL) | |
179 vm_exit_during_initialization("Failed to allocate " | |
180 "CM region mark stack"); | |
181 _index = 0; | |
182 // QQQQ cast ... | |
183 _capacity = (jint) size; | |
184 _oops_do_bound = -1; | |
185 NOT_PRODUCT(_max_depth = 0); | |
186 } | |
187 | |
188 CMMarkStack::~CMMarkStack() { | |
189 if (_base != NULL) FREE_C_HEAP_ARRAY(oop, _base); | |
190 } | |
191 | |
192 void CMMarkStack::par_push(oop ptr) { | |
193 while (true) { | |
194 if (isFull()) { | |
195 _overflow = true; | |
196 return; | |
197 } | |
198 // Otherwise... | |
199 jint index = _index; | |
200 jint next_index = index+1; | |
201 jint res = Atomic::cmpxchg(next_index, &_index, index); | |
202 if (res == index) { | |
203 _base[index] = ptr; | |
204 // Note that we don't maintain this atomically. We could, but it | |
205 // doesn't seem necessary. | |
206 NOT_PRODUCT(_max_depth = MAX2(_max_depth, next_index)); | |
207 return; | |
208 } | |
209 // Otherwise, we need to try again. | |
210 } | |
211 } | |
212 | |
213 void CMMarkStack::par_adjoin_arr(oop* ptr_arr, int n) { | |
214 while (true) { | |
215 if (isFull()) { | |
216 _overflow = true; | |
217 return; | |
218 } | |
219 // Otherwise... | |
220 jint index = _index; | |
221 jint next_index = index + n; | |
222 if (next_index > _capacity) { | |
223 _overflow = true; | |
224 return; | |
225 } | |
226 jint res = Atomic::cmpxchg(next_index, &_index, index); | |
227 if (res == index) { | |
228 for (int i = 0; i < n; i++) { | |
229 int ind = index + i; | |
230 assert(ind < _capacity, "By overflow test above."); | |
231 _base[ind] = ptr_arr[i]; | |
232 } | |
233 NOT_PRODUCT(_max_depth = MAX2(_max_depth, next_index)); | |
234 return; | |
235 } | |
236 // Otherwise, we need to try again. | |
237 } | |
238 } | |
239 | |
240 | |
241 void CMMarkStack::par_push_arr(oop* ptr_arr, int n) { | |
242 MutexLockerEx x(ParGCRareEvent_lock, Mutex::_no_safepoint_check_flag); | |
243 jint start = _index; | |
244 jint next_index = start + n; | |
245 if (next_index > _capacity) { | |
246 _overflow = true; | |
247 return; | |
248 } | |
249 // Otherwise. | |
250 _index = next_index; | |
251 for (int i = 0; i < n; i++) { | |
252 int ind = start + i; | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
253 assert(ind < _capacity, "By overflow test above."); |
342 | 254 _base[ind] = ptr_arr[i]; |
255 } | |
256 } | |
257 | |
258 | |
259 bool CMMarkStack::par_pop_arr(oop* ptr_arr, int max, int* n) { | |
260 MutexLockerEx x(ParGCRareEvent_lock, Mutex::_no_safepoint_check_flag); | |
261 jint index = _index; | |
262 if (index == 0) { | |
263 *n = 0; | |
264 return false; | |
265 } else { | |
266 int k = MIN2(max, index); | |
267 jint new_ind = index - k; | |
268 for (int j = 0; j < k; j++) { | |
269 ptr_arr[j] = _base[new_ind + j]; | |
270 } | |
271 _index = new_ind; | |
272 *n = k; | |
273 return true; | |
274 } | |
275 } | |
276 | |
277 | |
278 CMRegionStack::CMRegionStack() : _base(NULL) {} | |
279 | |
280 void CMRegionStack::allocate(size_t size) { | |
281 _base = NEW_C_HEAP_ARRAY(MemRegion, size); | |
282 if (_base == NULL) | |
283 vm_exit_during_initialization("Failed to allocate " | |
284 "CM region mark stack"); | |
285 _index = 0; | |
286 // QQQQ cast ... | |
287 _capacity = (jint) size; | |
288 } | |
289 | |
290 CMRegionStack::~CMRegionStack() { | |
291 if (_base != NULL) FREE_C_HEAP_ARRAY(oop, _base); | |
292 } | |
293 | |
1835
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
294 void CMRegionStack::push_lock_free(MemRegion mr) { |
342 | 295 assert(mr.word_size() > 0, "Precondition"); |
296 while (true) { | |
1835
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
297 jint index = _index; |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
298 |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
299 if (index >= _capacity) { |
342 | 300 _overflow = true; |
301 return; | |
302 } | |
303 // Otherwise... | |
304 jint next_index = index+1; | |
305 jint res = Atomic::cmpxchg(next_index, &_index, index); | |
306 if (res == index) { | |
307 _base[index] = mr; | |
308 return; | |
309 } | |
310 // Otherwise, we need to try again. | |
311 } | |
312 } | |
313 | |
1835
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
314 // Lock-free pop of the region stack. Called during the concurrent |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
315 // marking / remark phases. Should only be called in tandem with |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
316 // other lock-free pops. |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
317 MemRegion CMRegionStack::pop_lock_free() { |
342 | 318 while (true) { |
319 jint index = _index; | |
320 | |
321 if (index == 0) { | |
322 return MemRegion(); | |
323 } | |
1835
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
324 // Otherwise... |
342 | 325 jint next_index = index-1; |
326 jint res = Atomic::cmpxchg(next_index, &_index, index); | |
327 if (res == index) { | |
328 MemRegion mr = _base[next_index]; | |
329 if (mr.start() != NULL) { | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
330 assert(mr.end() != NULL, "invariant"); |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
331 assert(mr.word_size() > 0, "invariant"); |
342 | 332 return mr; |
333 } else { | |
334 // that entry was invalidated... let's skip it | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
335 assert(mr.end() == NULL, "invariant"); |
342 | 336 } |
337 } | |
338 // Otherwise, we need to try again. | |
339 } | |
340 } | |
1835
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
341 |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
342 #if 0 |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
343 // The routines that manipulate the region stack with a lock are |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
344 // not currently used. They should be retained, however, as a |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
345 // diagnostic aid. |
1358
72f725c5a7be
6940310: G1: MT-unsafe calls to CM::region_stack_push() / CM::region_stack_pop()
tonyp
parents:
1317
diff
changeset
|
346 |
72f725c5a7be
6940310: G1: MT-unsafe calls to CM::region_stack_push() / CM::region_stack_pop()
tonyp
parents:
1317
diff
changeset
|
347 void CMRegionStack::push_with_lock(MemRegion mr) { |
72f725c5a7be
6940310: G1: MT-unsafe calls to CM::region_stack_push() / CM::region_stack_pop()
tonyp
parents:
1317
diff
changeset
|
348 assert(mr.word_size() > 0, "Precondition"); |
72f725c5a7be
6940310: G1: MT-unsafe calls to CM::region_stack_push() / CM::region_stack_pop()
tonyp
parents:
1317
diff
changeset
|
349 MutexLockerEx x(CMRegionStack_lock, Mutex::_no_safepoint_check_flag); |
72f725c5a7be
6940310: G1: MT-unsafe calls to CM::region_stack_push() / CM::region_stack_pop()
tonyp
parents:
1317
diff
changeset
|
350 |
72f725c5a7be
6940310: G1: MT-unsafe calls to CM::region_stack_push() / CM::region_stack_pop()
tonyp
parents:
1317
diff
changeset
|
351 if (isFull()) { |
72f725c5a7be
6940310: G1: MT-unsafe calls to CM::region_stack_push() / CM::region_stack_pop()
tonyp
parents:
1317
diff
changeset
|
352 _overflow = true; |
72f725c5a7be
6940310: G1: MT-unsafe calls to CM::region_stack_push() / CM::region_stack_pop()
tonyp
parents:
1317
diff
changeset
|
353 return; |
72f725c5a7be
6940310: G1: MT-unsafe calls to CM::region_stack_push() / CM::region_stack_pop()
tonyp
parents:
1317
diff
changeset
|
354 } |
72f725c5a7be
6940310: G1: MT-unsafe calls to CM::region_stack_push() / CM::region_stack_pop()
tonyp
parents:
1317
diff
changeset
|
355 |
72f725c5a7be
6940310: G1: MT-unsafe calls to CM::region_stack_push() / CM::region_stack_pop()
tonyp
parents:
1317
diff
changeset
|
356 _base[_index] = mr; |
72f725c5a7be
6940310: G1: MT-unsafe calls to CM::region_stack_push() / CM::region_stack_pop()
tonyp
parents:
1317
diff
changeset
|
357 _index += 1; |
72f725c5a7be
6940310: G1: MT-unsafe calls to CM::region_stack_push() / CM::region_stack_pop()
tonyp
parents:
1317
diff
changeset
|
358 } |
72f725c5a7be
6940310: G1: MT-unsafe calls to CM::region_stack_push() / CM::region_stack_pop()
tonyp
parents:
1317
diff
changeset
|
359 |
72f725c5a7be
6940310: G1: MT-unsafe calls to CM::region_stack_push() / CM::region_stack_pop()
tonyp
parents:
1317
diff
changeset
|
360 MemRegion CMRegionStack::pop_with_lock() { |
72f725c5a7be
6940310: G1: MT-unsafe calls to CM::region_stack_push() / CM::region_stack_pop()
tonyp
parents:
1317
diff
changeset
|
361 MutexLockerEx x(CMRegionStack_lock, Mutex::_no_safepoint_check_flag); |
72f725c5a7be
6940310: G1: MT-unsafe calls to CM::region_stack_push() / CM::region_stack_pop()
tonyp
parents:
1317
diff
changeset
|
362 |
72f725c5a7be
6940310: G1: MT-unsafe calls to CM::region_stack_push() / CM::region_stack_pop()
tonyp
parents:
1317
diff
changeset
|
363 while (true) { |
72f725c5a7be
6940310: G1: MT-unsafe calls to CM::region_stack_push() / CM::region_stack_pop()
tonyp
parents:
1317
diff
changeset
|
364 if (_index == 0) { |
72f725c5a7be
6940310: G1: MT-unsafe calls to CM::region_stack_push() / CM::region_stack_pop()
tonyp
parents:
1317
diff
changeset
|
365 return MemRegion(); |
72f725c5a7be
6940310: G1: MT-unsafe calls to CM::region_stack_push() / CM::region_stack_pop()
tonyp
parents:
1317
diff
changeset
|
366 } |
72f725c5a7be
6940310: G1: MT-unsafe calls to CM::region_stack_push() / CM::region_stack_pop()
tonyp
parents:
1317
diff
changeset
|
367 _index -= 1; |
72f725c5a7be
6940310: G1: MT-unsafe calls to CM::region_stack_push() / CM::region_stack_pop()
tonyp
parents:
1317
diff
changeset
|
368 |
72f725c5a7be
6940310: G1: MT-unsafe calls to CM::region_stack_push() / CM::region_stack_pop()
tonyp
parents:
1317
diff
changeset
|
369 MemRegion mr = _base[_index]; |
72f725c5a7be
6940310: G1: MT-unsafe calls to CM::region_stack_push() / CM::region_stack_pop()
tonyp
parents:
1317
diff
changeset
|
370 if (mr.start() != NULL) { |
72f725c5a7be
6940310: G1: MT-unsafe calls to CM::region_stack_push() / CM::region_stack_pop()
tonyp
parents:
1317
diff
changeset
|
371 assert(mr.end() != NULL, "invariant"); |
72f725c5a7be
6940310: G1: MT-unsafe calls to CM::region_stack_push() / CM::region_stack_pop()
tonyp
parents:
1317
diff
changeset
|
372 assert(mr.word_size() > 0, "invariant"); |
72f725c5a7be
6940310: G1: MT-unsafe calls to CM::region_stack_push() / CM::region_stack_pop()
tonyp
parents:
1317
diff
changeset
|
373 return mr; |
72f725c5a7be
6940310: G1: MT-unsafe calls to CM::region_stack_push() / CM::region_stack_pop()
tonyp
parents:
1317
diff
changeset
|
374 } else { |
72f725c5a7be
6940310: G1: MT-unsafe calls to CM::region_stack_push() / CM::region_stack_pop()
tonyp
parents:
1317
diff
changeset
|
375 // that entry was invalidated... let's skip it |
72f725c5a7be
6940310: G1: MT-unsafe calls to CM::region_stack_push() / CM::region_stack_pop()
tonyp
parents:
1317
diff
changeset
|
376 assert(mr.end() == NULL, "invariant"); |
72f725c5a7be
6940310: G1: MT-unsafe calls to CM::region_stack_push() / CM::region_stack_pop()
tonyp
parents:
1317
diff
changeset
|
377 } |
72f725c5a7be
6940310: G1: MT-unsafe calls to CM::region_stack_push() / CM::region_stack_pop()
tonyp
parents:
1317
diff
changeset
|
378 } |
72f725c5a7be
6940310: G1: MT-unsafe calls to CM::region_stack_push() / CM::region_stack_pop()
tonyp
parents:
1317
diff
changeset
|
379 } |
1835
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
380 #endif |
342 | 381 |
382 bool CMRegionStack::invalidate_entries_into_cset() { | |
383 bool result = false; | |
384 G1CollectedHeap* g1h = G1CollectedHeap::heap(); | |
385 for (int i = 0; i < _oops_do_bound; ++i) { | |
386 MemRegion mr = _base[i]; | |
387 if (mr.start() != NULL) { | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
388 assert(mr.end() != NULL, "invariant"); |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
389 assert(mr.word_size() > 0, "invariant"); |
342 | 390 HeapRegion* hr = g1h->heap_region_containing(mr.start()); |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
391 assert(hr != NULL, "invariant"); |
342 | 392 if (hr->in_collection_set()) { |
393 // The region points into the collection set | |
394 _base[i] = MemRegion(); | |
395 result = true; | |
396 } | |
397 } else { | |
398 // that entry was invalidated... let's skip it | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
399 assert(mr.end() == NULL, "invariant"); |
342 | 400 } |
401 } | |
402 return result; | |
403 } | |
404 | |
405 template<class OopClosureClass> | |
406 bool CMMarkStack::drain(OopClosureClass* cl, CMBitMap* bm, bool yield_after) { | |
407 assert(!_drain_in_progress || !_drain_in_progress_yields || yield_after | |
408 || SafepointSynchronize::is_at_safepoint(), | |
409 "Drain recursion must be yield-safe."); | |
410 bool res = true; | |
411 debug_only(_drain_in_progress = true); | |
412 debug_only(_drain_in_progress_yields = yield_after); | |
413 while (!isEmpty()) { | |
414 oop newOop = pop(); | |
415 assert(G1CollectedHeap::heap()->is_in_reserved(newOop), "Bad pop"); | |
416 assert(newOop->is_oop(), "Expected an oop"); | |
417 assert(bm == NULL || bm->isMarked((HeapWord*)newOop), | |
418 "only grey objects on this stack"); | |
419 // iterate over the oops in this oop, marking and pushing | |
420 // the ones in CMS generation. | |
421 newOop->oop_iterate(cl); | |
422 if (yield_after && _cm->do_yield_check()) { | |
423 res = false; break; | |
424 } | |
425 } | |
426 debug_only(_drain_in_progress = false); | |
427 return res; | |
428 } | |
429 | |
430 void CMMarkStack::oops_do(OopClosure* f) { | |
431 if (_index == 0) return; | |
432 assert(_oops_do_bound != -1 && _oops_do_bound <= _index, | |
433 "Bound must be set."); | |
434 for (int i = 0; i < _oops_do_bound; i++) { | |
435 f->do_oop(&_base[i]); | |
436 } | |
437 _oops_do_bound = -1; | |
438 } | |
439 | |
440 bool ConcurrentMark::not_yet_marked(oop obj) const { | |
441 return (_g1h->is_obj_ill(obj) | |
442 || (_g1h->is_in_permanent(obj) | |
443 && !nextMarkBitMap()->isMarked((HeapWord*)obj))); | |
444 } | |
445 | |
446 #ifdef _MSC_VER // the use of 'this' below gets a warning, make it go away | |
447 #pragma warning( disable:4355 ) // 'this' : used in base member initializer list | |
448 #endif // _MSC_VER | |
449 | |
450 ConcurrentMark::ConcurrentMark(ReservedSpace rs, | |
451 int max_regions) : | |
452 _markBitMap1(rs, MinObjAlignment - 1), | |
453 _markBitMap2(rs, MinObjAlignment - 1), | |
454 | |
455 _parallel_marking_threads(0), | |
456 _sleep_factor(0.0), | |
457 _marking_task_overhead(1.0), | |
458 _cleanup_sleep_factor(0.0), | |
459 _cleanup_task_overhead(1.0), | |
460 _region_bm(max_regions, false /* in_resource_area*/), | |
461 _card_bm((rs.size() + CardTableModRefBS::card_size - 1) >> | |
462 CardTableModRefBS::card_shift, | |
463 false /* in_resource_area*/), | |
464 _prevMarkBitMap(&_markBitMap1), | |
465 _nextMarkBitMap(&_markBitMap2), | |
466 _at_least_one_mark_complete(false), | |
467 | |
468 _markStack(this), | |
469 _regionStack(), | |
470 // _finger set in set_non_marking_state | |
471 | |
472 _max_task_num(MAX2(ParallelGCThreads, (size_t)1)), | |
473 // _active_tasks set in set_non_marking_state | |
474 // _tasks set inside the constructor | |
475 _task_queues(new CMTaskQueueSet((int) _max_task_num)), | |
476 _terminator(ParallelTaskTerminator((int) _max_task_num, _task_queues)), | |
477 | |
478 _has_overflown(false), | |
479 _concurrent(false), | |
619
7ea5ca260b28
6814467: G1: small fixes related to concurrent marking verboseness
tonyp
parents:
470
diff
changeset
|
480 _has_aborted(false), |
7ea5ca260b28
6814467: G1: small fixes related to concurrent marking verboseness
tonyp
parents:
470
diff
changeset
|
481 _restart_for_overflow(false), |
7ea5ca260b28
6814467: G1: small fixes related to concurrent marking verboseness
tonyp
parents:
470
diff
changeset
|
482 _concurrent_marking_in_progress(false), |
7ea5ca260b28
6814467: G1: small fixes related to concurrent marking verboseness
tonyp
parents:
470
diff
changeset
|
483 _should_gray_objects(false), |
342 | 484 |
485 // _verbose_level set below | |
486 | |
487 _init_times(), | |
488 _remark_times(), _remark_mark_times(), _remark_weak_ref_times(), | |
489 _cleanup_times(), | |
490 _total_counting_time(0.0), | |
491 _total_rs_scrub_time(0.0), | |
492 | |
936 | 493 _parallel_workers(NULL) |
342 | 494 { |
495 CMVerboseLevel verbose_level = | |
496 (CMVerboseLevel) G1MarkingVerboseLevel; | |
497 if (verbose_level < no_verbose) | |
498 verbose_level = no_verbose; | |
499 if (verbose_level > high_verbose) | |
500 verbose_level = high_verbose; | |
501 _verbose_level = verbose_level; | |
502 | |
503 if (verbose_low()) | |
504 gclog_or_tty->print_cr("[global] init, heap start = "PTR_FORMAT", " | |
505 "heap end = "PTR_FORMAT, _heap_start, _heap_end); | |
506 | |
1284 | 507 _markStack.allocate(MarkStackSize); |
751 | 508 _regionStack.allocate(G1MarkRegionStackSize); |
342 | 509 |
510 // Create & start a ConcurrentMark thread. | |
845
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
511 _cmThread = new ConcurrentMarkThread(this); |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
512 assert(cmThread() != NULL, "CM Thread should have been created"); |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
513 assert(cmThread()->cm() != NULL, "CM Thread should refer to this cm"); |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
514 |
342 | 515 _g1h = G1CollectedHeap::heap(); |
516 assert(CGC_lock != NULL, "Where's the CGC_lock?"); | |
517 assert(_markBitMap1.covers(rs), "_markBitMap1 inconsistency"); | |
518 assert(_markBitMap2.covers(rs), "_markBitMap2 inconsistency"); | |
519 | |
520 SATBMarkQueueSet& satb_qs = JavaThread::satb_mark_queue_set(); | |
1282 | 521 satb_qs.set_buffer_size(G1SATBBufferSize); |
342 | 522 |
523 int size = (int) MAX2(ParallelGCThreads, (size_t)1); | |
524 _par_cleanup_thread_state = NEW_C_HEAP_ARRAY(ParCleanupThreadState*, size); | |
525 for (int i = 0 ; i < size; i++) { | |
526 _par_cleanup_thread_state[i] = new ParCleanupThreadState; | |
527 } | |
528 | |
529 _tasks = NEW_C_HEAP_ARRAY(CMTask*, _max_task_num); | |
530 _accum_task_vtime = NEW_C_HEAP_ARRAY(double, _max_task_num); | |
531 | |
532 // so that the assertion in MarkingTaskQueue::task_queue doesn't fail | |
533 _active_tasks = _max_task_num; | |
534 for (int i = 0; i < (int) _max_task_num; ++i) { | |
535 CMTaskQueue* task_queue = new CMTaskQueue(); | |
536 task_queue->initialize(); | |
537 _task_queues->register_queue(i, task_queue); | |
538 | |
539 _tasks[i] = new CMTask(i, this, task_queue, _task_queues); | |
540 _accum_task_vtime[i] = 0.0; | |
541 } | |
542 | |
1284 | 543 if (ConcGCThreads > ParallelGCThreads) { |
544 vm_exit_during_initialization("Can't have more ConcGCThreads " | |
342 | 545 "than ParallelGCThreads."); |
546 } | |
547 if (ParallelGCThreads == 0) { | |
548 // if we are not running with any parallel GC threads we will not | |
549 // spawn any marking threads either | |
550 _parallel_marking_threads = 0; | |
551 _sleep_factor = 0.0; | |
552 _marking_task_overhead = 1.0; | |
553 } else { | |
1284 | 554 if (ConcGCThreads > 0) { |
555 // notice that ConcGCThreads overwrites G1MarkingOverheadPercent | |
342 | 556 // if both are set |
557 | |
1284 | 558 _parallel_marking_threads = ConcGCThreads; |
342 | 559 _sleep_factor = 0.0; |
560 _marking_task_overhead = 1.0; | |
751 | 561 } else if (G1MarkingOverheadPercent > 0) { |
342 | 562 // we will calculate the number of parallel marking threads |
563 // based on a target overhead with respect to the soft real-time | |
564 // goal | |
565 | |
751 | 566 double marking_overhead = (double) G1MarkingOverheadPercent / 100.0; |
342 | 567 double overall_cm_overhead = |
751 | 568 (double) MaxGCPauseMillis * marking_overhead / |
569 (double) GCPauseIntervalMillis; | |
342 | 570 double cpu_ratio = 1.0 / (double) os::processor_count(); |
571 double marking_thread_num = ceil(overall_cm_overhead / cpu_ratio); | |
572 double marking_task_overhead = | |
573 overall_cm_overhead / marking_thread_num * | |
574 (double) os::processor_count(); | |
575 double sleep_factor = | |
576 (1.0 - marking_task_overhead) / marking_task_overhead; | |
577 | |
578 _parallel_marking_threads = (size_t) marking_thread_num; | |
579 _sleep_factor = sleep_factor; | |
580 _marking_task_overhead = marking_task_overhead; | |
581 } else { | |
582 _parallel_marking_threads = MAX2((ParallelGCThreads + 2) / 4, (size_t)1); | |
583 _sleep_factor = 0.0; | |
584 _marking_task_overhead = 1.0; | |
585 } | |
586 | |
587 if (parallel_marking_threads() > 1) | |
588 _cleanup_task_overhead = 1.0; | |
589 else | |
590 _cleanup_task_overhead = marking_task_overhead(); | |
591 _cleanup_sleep_factor = | |
592 (1.0 - cleanup_task_overhead()) / cleanup_task_overhead(); | |
593 | |
594 #if 0 | |
595 gclog_or_tty->print_cr("Marking Threads %d", parallel_marking_threads()); | |
596 gclog_or_tty->print_cr("CM Marking Task Overhead %1.4lf", marking_task_overhead()); | |
597 gclog_or_tty->print_cr("CM Sleep Factor %1.4lf", sleep_factor()); | |
598 gclog_or_tty->print_cr("CL Marking Task Overhead %1.4lf", cleanup_task_overhead()); | |
599 gclog_or_tty->print_cr("CL Sleep Factor %1.4lf", cleanup_sleep_factor()); | |
600 #endif | |
601 | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
602 guarantee(parallel_marking_threads() > 0, "peace of mind"); |
1833
8b10f48633dc
6984287: Regularize how GC parallel workers are specified.
jmasa
parents:
1719
diff
changeset
|
603 _parallel_workers = new FlexibleWorkGang("G1 Parallel Marking Threads", |
8b10f48633dc
6984287: Regularize how GC parallel workers are specified.
jmasa
parents:
1719
diff
changeset
|
604 (int) _parallel_marking_threads, false, true); |
8b10f48633dc
6984287: Regularize how GC parallel workers are specified.
jmasa
parents:
1719
diff
changeset
|
605 if (_parallel_workers == NULL) { |
342 | 606 vm_exit_during_initialization("Failed necessary allocation."); |
1833
8b10f48633dc
6984287: Regularize how GC parallel workers are specified.
jmasa
parents:
1719
diff
changeset
|
607 } else { |
8b10f48633dc
6984287: Regularize how GC parallel workers are specified.
jmasa
parents:
1719
diff
changeset
|
608 _parallel_workers->initialize_workers(); |
8b10f48633dc
6984287: Regularize how GC parallel workers are specified.
jmasa
parents:
1719
diff
changeset
|
609 } |
342 | 610 } |
611 | |
612 // so that the call below can read a sensible value | |
613 _heap_start = (HeapWord*) rs.base(); | |
614 set_non_marking_state(); | |
615 } | |
616 | |
617 void ConcurrentMark::update_g1_committed(bool force) { | |
618 // If concurrent marking is not in progress, then we do not need to | |
619 // update _heap_end. This has a subtle and important | |
620 // side-effect. Imagine that two evacuation pauses happen between | |
621 // marking completion and remark. The first one can grow the | |
622 // heap (hence now the finger is below the heap end). Then, the | |
623 // second one could unnecessarily push regions on the region | |
624 // stack. This causes the invariant that the region stack is empty | |
625 // at the beginning of remark to be false. By ensuring that we do | |
626 // not observe heap expansions after marking is complete, then we do | |
627 // not have this problem. | |
628 if (!concurrent_marking_in_progress() && !force) | |
629 return; | |
630 | |
631 MemRegion committed = _g1h->g1_committed(); | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
632 assert(committed.start() == _heap_start, "start shouldn't change"); |
342 | 633 HeapWord* new_end = committed.end(); |
634 if (new_end > _heap_end) { | |
635 // The heap has been expanded. | |
636 | |
637 _heap_end = new_end; | |
638 } | |
639 // Notice that the heap can also shrink. However, this only happens | |
640 // during a Full GC (at least currently) and the entire marking | |
641 // phase will bail out and the task will not be restarted. So, let's | |
642 // do nothing. | |
643 } | |
644 | |
645 void ConcurrentMark::reset() { | |
646 // Starting values for these two. This should be called in a STW | |
647 // phase. CM will be notified of any future g1_committed expansions | |
648 // will be at the end of evacuation pauses, when tasks are | |
649 // inactive. | |
650 MemRegion committed = _g1h->g1_committed(); | |
651 _heap_start = committed.start(); | |
652 _heap_end = committed.end(); | |
653 | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
654 // Separated the asserts so that we know which one fires. |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
655 assert(_heap_start != NULL, "heap bounds should look ok"); |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
656 assert(_heap_end != NULL, "heap bounds should look ok"); |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
657 assert(_heap_start < _heap_end, "heap bounds should look ok"); |
342 | 658 |
659 // reset all the marking data structures and any necessary flags | |
660 clear_marking_state(); | |
661 | |
662 if (verbose_low()) | |
663 gclog_or_tty->print_cr("[global] resetting"); | |
664 | |
665 // We do reset all of them, since different phases will use | |
666 // different number of active threads. So, it's easiest to have all | |
667 // of them ready. | |
1835
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
668 for (int i = 0; i < (int) _max_task_num; ++i) { |
342 | 669 _tasks[i]->reset(_nextMarkBitMap); |
1835
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
670 } |
342 | 671 |
672 // we need this to make sure that the flag is on during the evac | |
673 // pause with initial mark piggy-backed | |
674 set_concurrent_marking_in_progress(); | |
675 } | |
676 | |
677 void ConcurrentMark::set_phase(size_t active_tasks, bool concurrent) { | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
678 assert(active_tasks <= _max_task_num, "we should not have more"); |
342 | 679 |
680 _active_tasks = active_tasks; | |
681 // Need to update the three data structures below according to the | |
682 // number of active threads for this phase. | |
683 _terminator = ParallelTaskTerminator((int) active_tasks, _task_queues); | |
684 _first_overflow_barrier_sync.set_n_workers((int) active_tasks); | |
685 _second_overflow_barrier_sync.set_n_workers((int) active_tasks); | |
686 | |
687 _concurrent = concurrent; | |
688 // We propagate this to all tasks, not just the active ones. | |
689 for (int i = 0; i < (int) _max_task_num; ++i) | |
690 _tasks[i]->set_concurrent(concurrent); | |
691 | |
692 if (concurrent) { | |
693 set_concurrent_marking_in_progress(); | |
694 } else { | |
695 // We currently assume that the concurrent flag has been set to | |
696 // false before we start remark. At this point we should also be | |
697 // in a STW phase. | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
698 assert(!concurrent_marking_in_progress(), "invariant"); |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
699 assert(_finger == _heap_end, "only way to get here"); |
342 | 700 update_g1_committed(true); |
701 } | |
702 } | |
703 | |
704 void ConcurrentMark::set_non_marking_state() { | |
705 // We set the global marking state to some default values when we're | |
706 // not doing marking. | |
707 clear_marking_state(); | |
708 _active_tasks = 0; | |
709 clear_concurrent_marking_in_progress(); | |
710 } | |
711 | |
712 ConcurrentMark::~ConcurrentMark() { | |
713 int size = (int) MAX2(ParallelGCThreads, (size_t)1); | |
714 for (int i = 0; i < size; i++) delete _par_cleanup_thread_state[i]; | |
715 FREE_C_HEAP_ARRAY(ParCleanupThreadState*, | |
716 _par_cleanup_thread_state); | |
717 | |
718 for (int i = 0; i < (int) _max_task_num; ++i) { | |
719 delete _task_queues->queue(i); | |
720 delete _tasks[i]; | |
721 } | |
722 delete _task_queues; | |
723 FREE_C_HEAP_ARRAY(CMTask*, _max_task_num); | |
724 } | |
725 | |
726 // This closure is used to mark refs into the g1 generation | |
727 // from external roots in the CMS bit map. | |
728 // Called at the first checkpoint. | |
729 // | |
730 | |
731 void ConcurrentMark::clearNextBitmap() { | |
1359
23b1b27ac76c
6909756: G1: guarantee(G1CollectedHeap::heap()->mark_in_progress(),"Precondition.")
tonyp
parents:
1358
diff
changeset
|
732 G1CollectedHeap* g1h = G1CollectedHeap::heap(); |
23b1b27ac76c
6909756: G1: guarantee(G1CollectedHeap::heap()->mark_in_progress(),"Precondition.")
tonyp
parents:
1358
diff
changeset
|
733 G1CollectorPolicy* g1p = g1h->g1_policy(); |
23b1b27ac76c
6909756: G1: guarantee(G1CollectedHeap::heap()->mark_in_progress(),"Precondition.")
tonyp
parents:
1358
diff
changeset
|
734 |
23b1b27ac76c
6909756: G1: guarantee(G1CollectedHeap::heap()->mark_in_progress(),"Precondition.")
tonyp
parents:
1358
diff
changeset
|
735 // Make sure that the concurrent mark thread looks to still be in |
23b1b27ac76c
6909756: G1: guarantee(G1CollectedHeap::heap()->mark_in_progress(),"Precondition.")
tonyp
parents:
1358
diff
changeset
|
736 // the current cycle. |
23b1b27ac76c
6909756: G1: guarantee(G1CollectedHeap::heap()->mark_in_progress(),"Precondition.")
tonyp
parents:
1358
diff
changeset
|
737 guarantee(cmThread()->during_cycle(), "invariant"); |
23b1b27ac76c
6909756: G1: guarantee(G1CollectedHeap::heap()->mark_in_progress(),"Precondition.")
tonyp
parents:
1358
diff
changeset
|
738 |
23b1b27ac76c
6909756: G1: guarantee(G1CollectedHeap::heap()->mark_in_progress(),"Precondition.")
tonyp
parents:
1358
diff
changeset
|
739 // We are finishing up the current cycle by clearing the next |
23b1b27ac76c
6909756: G1: guarantee(G1CollectedHeap::heap()->mark_in_progress(),"Precondition.")
tonyp
parents:
1358
diff
changeset
|
740 // marking bitmap and getting it ready for the next cycle. During |
23b1b27ac76c
6909756: G1: guarantee(G1CollectedHeap::heap()->mark_in_progress(),"Precondition.")
tonyp
parents:
1358
diff
changeset
|
741 // this time no other cycle can start. So, let's make sure that this |
23b1b27ac76c
6909756: G1: guarantee(G1CollectedHeap::heap()->mark_in_progress(),"Precondition.")
tonyp
parents:
1358
diff
changeset
|
742 // is the case. |
23b1b27ac76c
6909756: G1: guarantee(G1CollectedHeap::heap()->mark_in_progress(),"Precondition.")
tonyp
parents:
1358
diff
changeset
|
743 guarantee(!g1h->mark_in_progress(), "invariant"); |
23b1b27ac76c
6909756: G1: guarantee(G1CollectedHeap::heap()->mark_in_progress(),"Precondition.")
tonyp
parents:
1358
diff
changeset
|
744 |
23b1b27ac76c
6909756: G1: guarantee(G1CollectedHeap::heap()->mark_in_progress(),"Precondition.")
tonyp
parents:
1358
diff
changeset
|
745 // clear the mark bitmap (no grey objects to start with). |
23b1b27ac76c
6909756: G1: guarantee(G1CollectedHeap::heap()->mark_in_progress(),"Precondition.")
tonyp
parents:
1358
diff
changeset
|
746 // We need to do this in chunks and offer to yield in between |
23b1b27ac76c
6909756: G1: guarantee(G1CollectedHeap::heap()->mark_in_progress(),"Precondition.")
tonyp
parents:
1358
diff
changeset
|
747 // each chunk. |
23b1b27ac76c
6909756: G1: guarantee(G1CollectedHeap::heap()->mark_in_progress(),"Precondition.")
tonyp
parents:
1358
diff
changeset
|
748 HeapWord* start = _nextMarkBitMap->startWord(); |
23b1b27ac76c
6909756: G1: guarantee(G1CollectedHeap::heap()->mark_in_progress(),"Precondition.")
tonyp
parents:
1358
diff
changeset
|
749 HeapWord* end = _nextMarkBitMap->endWord(); |
23b1b27ac76c
6909756: G1: guarantee(G1CollectedHeap::heap()->mark_in_progress(),"Precondition.")
tonyp
parents:
1358
diff
changeset
|
750 HeapWord* cur = start; |
23b1b27ac76c
6909756: G1: guarantee(G1CollectedHeap::heap()->mark_in_progress(),"Precondition.")
tonyp
parents:
1358
diff
changeset
|
751 size_t chunkSize = M; |
23b1b27ac76c
6909756: G1: guarantee(G1CollectedHeap::heap()->mark_in_progress(),"Precondition.")
tonyp
parents:
1358
diff
changeset
|
752 while (cur < end) { |
23b1b27ac76c
6909756: G1: guarantee(G1CollectedHeap::heap()->mark_in_progress(),"Precondition.")
tonyp
parents:
1358
diff
changeset
|
753 HeapWord* next = cur + chunkSize; |
23b1b27ac76c
6909756: G1: guarantee(G1CollectedHeap::heap()->mark_in_progress(),"Precondition.")
tonyp
parents:
1358
diff
changeset
|
754 if (next > end) |
23b1b27ac76c
6909756: G1: guarantee(G1CollectedHeap::heap()->mark_in_progress(),"Precondition.")
tonyp
parents:
1358
diff
changeset
|
755 next = end; |
23b1b27ac76c
6909756: G1: guarantee(G1CollectedHeap::heap()->mark_in_progress(),"Precondition.")
tonyp
parents:
1358
diff
changeset
|
756 MemRegion mr(cur,next); |
23b1b27ac76c
6909756: G1: guarantee(G1CollectedHeap::heap()->mark_in_progress(),"Precondition.")
tonyp
parents:
1358
diff
changeset
|
757 _nextMarkBitMap->clearRange(mr); |
23b1b27ac76c
6909756: G1: guarantee(G1CollectedHeap::heap()->mark_in_progress(),"Precondition.")
tonyp
parents:
1358
diff
changeset
|
758 cur = next; |
23b1b27ac76c
6909756: G1: guarantee(G1CollectedHeap::heap()->mark_in_progress(),"Precondition.")
tonyp
parents:
1358
diff
changeset
|
759 do_yield_check(); |
23b1b27ac76c
6909756: G1: guarantee(G1CollectedHeap::heap()->mark_in_progress(),"Precondition.")
tonyp
parents:
1358
diff
changeset
|
760 |
23b1b27ac76c
6909756: G1: guarantee(G1CollectedHeap::heap()->mark_in_progress(),"Precondition.")
tonyp
parents:
1358
diff
changeset
|
761 // Repeat the asserts from above. We'll do them as asserts here to |
23b1b27ac76c
6909756: G1: guarantee(G1CollectedHeap::heap()->mark_in_progress(),"Precondition.")
tonyp
parents:
1358
diff
changeset
|
762 // minimize their overhead on the product. However, we'll have |
23b1b27ac76c
6909756: G1: guarantee(G1CollectedHeap::heap()->mark_in_progress(),"Precondition.")
tonyp
parents:
1358
diff
changeset
|
763 // them as guarantees at the beginning / end of the bitmap |
23b1b27ac76c
6909756: G1: guarantee(G1CollectedHeap::heap()->mark_in_progress(),"Precondition.")
tonyp
parents:
1358
diff
changeset
|
764 // clearing to get some checking in the product. |
23b1b27ac76c
6909756: G1: guarantee(G1CollectedHeap::heap()->mark_in_progress(),"Precondition.")
tonyp
parents:
1358
diff
changeset
|
765 assert(cmThread()->during_cycle(), "invariant"); |
23b1b27ac76c
6909756: G1: guarantee(G1CollectedHeap::heap()->mark_in_progress(),"Precondition.")
tonyp
parents:
1358
diff
changeset
|
766 assert(!g1h->mark_in_progress(), "invariant"); |
23b1b27ac76c
6909756: G1: guarantee(G1CollectedHeap::heap()->mark_in_progress(),"Precondition.")
tonyp
parents:
1358
diff
changeset
|
767 } |
23b1b27ac76c
6909756: G1: guarantee(G1CollectedHeap::heap()->mark_in_progress(),"Precondition.")
tonyp
parents:
1358
diff
changeset
|
768 |
23b1b27ac76c
6909756: G1: guarantee(G1CollectedHeap::heap()->mark_in_progress(),"Precondition.")
tonyp
parents:
1358
diff
changeset
|
769 // Repeat the asserts from above. |
23b1b27ac76c
6909756: G1: guarantee(G1CollectedHeap::heap()->mark_in_progress(),"Precondition.")
tonyp
parents:
1358
diff
changeset
|
770 guarantee(cmThread()->during_cycle(), "invariant"); |
23b1b27ac76c
6909756: G1: guarantee(G1CollectedHeap::heap()->mark_in_progress(),"Precondition.")
tonyp
parents:
1358
diff
changeset
|
771 guarantee(!g1h->mark_in_progress(), "invariant"); |
342 | 772 } |
773 | |
774 class NoteStartOfMarkHRClosure: public HeapRegionClosure { | |
775 public: | |
776 bool doHeapRegion(HeapRegion* r) { | |
777 if (!r->continuesHumongous()) { | |
778 r->note_start_of_marking(true); | |
779 } | |
780 return false; | |
781 } | |
782 }; | |
783 | |
784 void ConcurrentMark::checkpointRootsInitialPre() { | |
785 G1CollectedHeap* g1h = G1CollectedHeap::heap(); | |
786 G1CollectorPolicy* g1p = g1h->g1_policy(); | |
787 | |
788 _has_aborted = false; | |
789 | |
1547
fb1a39993f69
6951319: enable solaris builds using Sun Studio 12 update 1
jcoomes
parents:
1388
diff
changeset
|
790 #ifndef PRODUCT |
1044 | 791 if (G1PrintReachableAtInitialMark) { |
1388 | 792 print_reachable("at-cycle-start", |
793 true /* use_prev_marking */, true /* all */); | |
1044 | 794 } |
1547
fb1a39993f69
6951319: enable solaris builds using Sun Studio 12 update 1
jcoomes
parents:
1388
diff
changeset
|
795 #endif |
342 | 796 |
797 // Initialise marking structures. This has to be done in a STW phase. | |
798 reset(); | |
799 } | |
800 | |
801 class CMMarkRootsClosure: public OopsInGenClosure { | |
802 private: | |
803 ConcurrentMark* _cm; | |
804 G1CollectedHeap* _g1h; | |
805 bool _do_barrier; | |
806 | |
807 public: | |
808 CMMarkRootsClosure(ConcurrentMark* cm, | |
809 G1CollectedHeap* g1h, | |
810 bool do_barrier) : _cm(cm), _g1h(g1h), | |
811 _do_barrier(do_barrier) { } | |
812 | |
845
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
813 virtual void do_oop(narrowOop* p) { do_oop_work(p); } |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
814 virtual void do_oop( oop* p) { do_oop_work(p); } |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
815 |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
816 template <class T> void do_oop_work(T* p) { |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
817 T heap_oop = oopDesc::load_heap_oop(p); |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
818 if (!oopDesc::is_null(heap_oop)) { |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
819 oop obj = oopDesc::decode_heap_oop_not_null(heap_oop); |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
820 assert(obj->is_oop() || obj->mark() == NULL, |
342 | 821 "expected an oop, possibly with mark word displaced"); |
845
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
822 HeapWord* addr = (HeapWord*)obj; |
342 | 823 if (_g1h->is_in_g1_reserved(addr)) { |
845
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
824 _cm->grayRoot(obj); |
342 | 825 } |
826 } | |
827 if (_do_barrier) { | |
828 assert(!_g1h->is_in_g1_reserved(p), | |
829 "Should be called on external roots"); | |
830 do_barrier(p); | |
831 } | |
832 } | |
833 }; | |
834 | |
835 void ConcurrentMark::checkpointRootsInitialPost() { | |
836 G1CollectedHeap* g1h = G1CollectedHeap::heap(); | |
837 | |
838 // For each region note start of marking. | |
839 NoteStartOfMarkHRClosure startcl; | |
840 g1h->heap_region_iterate(&startcl); | |
841 | |
842 // Start weak-reference discovery. | |
843 ReferenceProcessor* rp = g1h->ref_processor(); | |
844 rp->verify_no_references_recorded(); | |
845 rp->enable_discovery(); // enable ("weak") refs discovery | |
457
27a80744a83b
6778647: snap(), snap_policy() should be renamed setup(), setup_policy()
ysr
parents:
453
diff
changeset
|
846 rp->setup_policy(false); // snapshot the soft ref policy to be used in this cycle |
342 | 847 |
848 SATBMarkQueueSet& satb_mq_set = JavaThread::satb_mark_queue_set(); | |
1317
d4197f8d516a
6935821: G1: threads created during marking do not active their SATB queues
tonyp
parents:
1314
diff
changeset
|
849 // This is the start of the marking cycle, we're expected all |
d4197f8d516a
6935821: G1: threads created during marking do not active their SATB queues
tonyp
parents:
1314
diff
changeset
|
850 // threads to have SATB queues with active set to false. |
d4197f8d516a
6935821: G1: threads created during marking do not active their SATB queues
tonyp
parents:
1314
diff
changeset
|
851 satb_mq_set.set_active_all_threads(true, /* new active value */ |
d4197f8d516a
6935821: G1: threads created during marking do not active their SATB queues
tonyp
parents:
1314
diff
changeset
|
852 false /* expected_active */); |
342 | 853 |
854 // update_g1_committed() will be called at the end of an evac pause | |
855 // when marking is on. So, it's also called at the end of the | |
856 // initial-mark pause to update the heap end, if the heap expands | |
857 // during it. No need to call it here. | |
858 } | |
859 | |
860 // Checkpoint the roots into this generation from outside | |
861 // this generation. [Note this initial checkpoint need only | |
862 // be approximate -- we'll do a catch up phase subsequently.] | |
863 void ConcurrentMark::checkpointRootsInitial() { | |
864 assert(SafepointSynchronize::is_at_safepoint(), "world should be stopped"); | |
865 G1CollectedHeap* g1h = G1CollectedHeap::heap(); | |
866 | |
867 double start = os::elapsedTime(); | |
868 | |
869 G1CollectorPolicy* g1p = G1CollectedHeap::heap()->g1_policy(); | |
870 g1p->record_concurrent_mark_init_start(); | |
871 checkpointRootsInitialPre(); | |
872 | |
873 // YSR: when concurrent precleaning is in place, we'll | |
874 // need to clear the cached card table here | |
875 | |
876 ResourceMark rm; | |
877 HandleMark hm; | |
878 | |
879 g1h->ensure_parsability(false); | |
880 g1h->perm_gen()->save_marks(); | |
881 | |
882 CMMarkRootsClosure notOlder(this, g1h, false); | |
883 CMMarkRootsClosure older(this, g1h, true); | |
884 | |
885 g1h->set_marking_started(); | |
886 g1h->rem_set()->prepare_for_younger_refs_iterate(false); | |
887 | |
989
148e5441d916
6863023: need non-perm oops in code cache for JSR 292
jrose
parents:
912
diff
changeset
|
888 g1h->process_strong_roots(true, // activate StrongRootsScope |
148e5441d916
6863023: need non-perm oops in code cache for JSR 292
jrose
parents:
912
diff
changeset
|
889 false, // fake perm gen collection |
342 | 890 SharedHeap::SO_AllClasses, |
891 ¬Older, // Regular roots | |
989
148e5441d916
6863023: need non-perm oops in code cache for JSR 292
jrose
parents:
912
diff
changeset
|
892 NULL, // do not visit active blobs |
342 | 893 &older // Perm Gen Roots |
894 ); | |
895 checkpointRootsInitialPost(); | |
896 | |
897 // Statistics. | |
898 double end = os::elapsedTime(); | |
899 _init_times.add((end - start) * 1000.0); | |
900 | |
901 g1p->record_concurrent_mark_init_end(); | |
902 } | |
903 | |
904 /* | |
905 Notice that in the next two methods, we actually leave the STS | |
906 during the barrier sync and join it immediately afterwards. If we | |
907 do not do this, this then the following deadlock can occur: one | |
908 thread could be in the barrier sync code, waiting for the other | |
909 thread to also sync up, whereas another one could be trying to | |
910 yield, while also waiting for the other threads to sync up too. | |
911 | |
912 Because the thread that does the sync barrier has left the STS, it | |
913 is possible to be suspended for a Full GC or an evacuation pause | |
914 could occur. This is actually safe, since the entering the sync | |
915 barrier is one of the last things do_marking_step() does, and it | |
916 doesn't manipulate any data structures afterwards. | |
917 */ | |
918 | |
919 void ConcurrentMark::enter_first_sync_barrier(int task_num) { | |
920 if (verbose_low()) | |
921 gclog_or_tty->print_cr("[%d] entering first barrier", task_num); | |
922 | |
923 ConcurrentGCThread::stsLeave(); | |
924 _first_overflow_barrier_sync.enter(); | |
925 ConcurrentGCThread::stsJoin(); | |
926 // at this point everyone should have synced up and not be doing any | |
927 // more work | |
928 | |
929 if (verbose_low()) | |
930 gclog_or_tty->print_cr("[%d] leaving first barrier", task_num); | |
931 | |
932 // let task 0 do this | |
933 if (task_num == 0) { | |
934 // task 0 is responsible for clearing the global data structures | |
935 clear_marking_state(); | |
936 | |
937 if (PrintGC) { | |
938 gclog_or_tty->date_stamp(PrintGCDateStamps); | |
939 gclog_or_tty->stamp(PrintGCTimeStamps); | |
940 gclog_or_tty->print_cr("[GC concurrent-mark-reset-for-overflow]"); | |
941 } | |
942 } | |
943 | |
944 // after this, each task should reset its own data structures then | |
945 // then go into the second barrier | |
946 } | |
947 | |
948 void ConcurrentMark::enter_second_sync_barrier(int task_num) { | |
949 if (verbose_low()) | |
950 gclog_or_tty->print_cr("[%d] entering second barrier", task_num); | |
951 | |
952 ConcurrentGCThread::stsLeave(); | |
953 _second_overflow_barrier_sync.enter(); | |
954 ConcurrentGCThread::stsJoin(); | |
955 // at this point everything should be re-initialised and ready to go | |
956 | |
957 if (verbose_low()) | |
958 gclog_or_tty->print_cr("[%d] leaving second barrier", task_num); | |
959 } | |
960 | |
961 void ConcurrentMark::grayRoot(oop p) { | |
962 HeapWord* addr = (HeapWord*) p; | |
963 // We can't really check against _heap_start and _heap_end, since it | |
964 // is possible during an evacuation pause with piggy-backed | |
965 // initial-mark that the committed space is expanded during the | |
966 // pause without CM observing this change. So the assertions below | |
967 // is a bit conservative; but better than nothing. | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
968 assert(_g1h->g1_committed().contains(addr), |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
969 "address should be within the heap bounds"); |
342 | 970 |
971 if (!_nextMarkBitMap->isMarked(addr)) | |
972 _nextMarkBitMap->parMark(addr); | |
973 } | |
974 | |
975 void ConcurrentMark::grayRegionIfNecessary(MemRegion mr) { | |
976 // The objects on the region have already been marked "in bulk" by | |
977 // the caller. We only need to decide whether to push the region on | |
978 // the region stack or not. | |
979 | |
980 if (!concurrent_marking_in_progress() || !_should_gray_objects) | |
981 // We're done with marking and waiting for remark. We do not need to | |
982 // push anything else on the region stack. | |
983 return; | |
984 | |
985 HeapWord* finger = _finger; | |
986 | |
987 if (verbose_low()) | |
988 gclog_or_tty->print_cr("[global] attempting to push " | |
989 "region ["PTR_FORMAT", "PTR_FORMAT"), finger is at " | |
990 PTR_FORMAT, mr.start(), mr.end(), finger); | |
991 | |
992 if (mr.start() < finger) { | |
993 // The finger is always heap region aligned and it is not possible | |
994 // for mr to span heap regions. | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
995 assert(mr.end() <= finger, "invariant"); |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
996 |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
997 // Separated the asserts so that we know which one fires. |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
998 assert(mr.start() <= mr.end(), |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
999 "region boundaries should fall within the committed space"); |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
1000 assert(_heap_start <= mr.start(), |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
1001 "region boundaries should fall within the committed space"); |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
1002 assert(mr.end() <= _heap_end, |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
1003 "region boundaries should fall within the committed space"); |
342 | 1004 if (verbose_low()) |
1005 gclog_or_tty->print_cr("[global] region ["PTR_FORMAT", "PTR_FORMAT") " | |
1006 "below the finger, pushing it", | |
1007 mr.start(), mr.end()); | |
1008 | |
1835
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
1009 if (!region_stack_push_lock_free(mr)) { |
342 | 1010 if (verbose_low()) |
1011 gclog_or_tty->print_cr("[global] region stack has overflown."); | |
1012 } | |
1013 } | |
1014 } | |
1015 | |
1016 void ConcurrentMark::markAndGrayObjectIfNecessary(oop p) { | |
1017 // The object is not marked by the caller. We need to at least mark | |
1018 // it and maybe push in on the stack. | |
1019 | |
1020 HeapWord* addr = (HeapWord*)p; | |
1021 if (!_nextMarkBitMap->isMarked(addr)) { | |
1022 // We definitely need to mark it, irrespective whether we bail out | |
1023 // because we're done with marking. | |
1024 if (_nextMarkBitMap->parMark(addr)) { | |
1025 if (!concurrent_marking_in_progress() || !_should_gray_objects) | |
1026 // If we're done with concurrent marking and we're waiting for | |
1027 // remark, then we're not pushing anything on the stack. | |
1028 return; | |
1029 | |
1030 // No OrderAccess:store_load() is needed. It is implicit in the | |
1031 // CAS done in parMark(addr) above | |
1032 HeapWord* finger = _finger; | |
1033 | |
1034 if (addr < finger) { | |
1035 if (!mark_stack_push(oop(addr))) { | |
1036 if (verbose_low()) | |
1037 gclog_or_tty->print_cr("[global] global stack overflow " | |
1038 "during parMark"); | |
1039 } | |
1040 } | |
1041 } | |
1042 } | |
1043 } | |
1044 | |
1045 class CMConcurrentMarkingTask: public AbstractGangTask { | |
1046 private: | |
1047 ConcurrentMark* _cm; | |
1048 ConcurrentMarkThread* _cmt; | |
1049 | |
1050 public: | |
1051 void work(int worker_i) { | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
1052 assert(Thread::current()->is_ConcurrentGC_thread(), |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
1053 "this should only be done by a conc GC thread"); |
1974
fd1d227ef1b9
6983204: G1: Nightly test nsk/regression/b4958615 failing with +ExplicitGCInvokesConcurrent
johnc
parents:
1972
diff
changeset
|
1054 ResourceMark rm; |
342 | 1055 |
1056 double start_vtime = os::elapsedVTime(); | |
1057 | |
1058 ConcurrentGCThread::stsJoin(); | |
1059 | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
1060 assert((size_t) worker_i < _cm->active_tasks(), "invariant"); |
342 | 1061 CMTask* the_task = _cm->task(worker_i); |
1062 the_task->record_start_time(); | |
1063 if (!_cm->has_aborted()) { | |
1064 do { | |
1065 double start_vtime_sec = os::elapsedVTime(); | |
1066 double start_time_sec = os::elapsedTime(); | |
1067 the_task->do_marking_step(10.0); | |
1068 double end_time_sec = os::elapsedTime(); | |
1069 double end_vtime_sec = os::elapsedVTime(); | |
1070 double elapsed_vtime_sec = end_vtime_sec - start_vtime_sec; | |
1071 double elapsed_time_sec = end_time_sec - start_time_sec; | |
1072 _cm->clear_has_overflown(); | |
1073 | |
1074 bool ret = _cm->do_yield_check(worker_i); | |
1075 | |
1076 jlong sleep_time_ms; | |
1077 if (!_cm->has_aborted() && the_task->has_aborted()) { | |
1078 sleep_time_ms = | |
1079 (jlong) (elapsed_vtime_sec * _cm->sleep_factor() * 1000.0); | |
1080 ConcurrentGCThread::stsLeave(); | |
1081 os::sleep(Thread::current(), sleep_time_ms, false); | |
1082 ConcurrentGCThread::stsJoin(); | |
1083 } | |
1084 double end_time2_sec = os::elapsedTime(); | |
1085 double elapsed_time2_sec = end_time2_sec - start_time_sec; | |
1086 | |
1087 #if 0 | |
1088 gclog_or_tty->print_cr("CM: elapsed %1.4lf ms, sleep %1.4lf ms, " | |
1089 "overhead %1.4lf", | |
1090 elapsed_vtime_sec * 1000.0, (double) sleep_time_ms, | |
1091 the_task->conc_overhead(os::elapsedTime()) * 8.0); | |
1092 gclog_or_tty->print_cr("elapsed time %1.4lf ms, time 2: %1.4lf ms", | |
1093 elapsed_time_sec * 1000.0, elapsed_time2_sec * 1000.0); | |
1094 #endif | |
1095 } while (!_cm->has_aborted() && the_task->has_aborted()); | |
1096 } | |
1097 the_task->record_end_time(); | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
1098 guarantee(!the_task->has_aborted() || _cm->has_aborted(), "invariant"); |
342 | 1099 |
1100 ConcurrentGCThread::stsLeave(); | |
1101 | |
1102 double end_vtime = os::elapsedVTime(); | |
1103 _cm->update_accum_task_vtime(worker_i, end_vtime - start_vtime); | |
1104 } | |
1105 | |
1106 CMConcurrentMarkingTask(ConcurrentMark* cm, | |
1107 ConcurrentMarkThread* cmt) : | |
1108 AbstractGangTask("Concurrent Mark"), _cm(cm), _cmt(cmt) { } | |
1109 | |
1110 ~CMConcurrentMarkingTask() { } | |
1111 }; | |
1112 | |
1113 void ConcurrentMark::markFromRoots() { | |
1114 // we might be tempted to assert that: | |
1115 // assert(asynch == !SafepointSynchronize::is_at_safepoint(), | |
1116 // "inconsistent argument?"); | |
1117 // However that wouldn't be right, because it's possible that | |
1118 // a safepoint is indeed in progress as a younger generation | |
1119 // stop-the-world GC happens even as we mark in this generation. | |
1120 | |
1121 _restart_for_overflow = false; | |
1122 | |
1123 set_phase(MAX2((size_t) 1, parallel_marking_threads()), true); | |
1124 | |
1125 CMConcurrentMarkingTask markingTask(this, cmThread()); | |
1126 if (parallel_marking_threads() > 0) | |
1127 _parallel_workers->run_task(&markingTask); | |
1128 else | |
1129 markingTask.work(0); | |
1130 print_stats(); | |
1131 } | |
1132 | |
1133 void ConcurrentMark::checkpointRootsFinal(bool clear_all_soft_refs) { | |
1134 // world is stopped at this checkpoint | |
1135 assert(SafepointSynchronize::is_at_safepoint(), | |
1136 "world should be stopped"); | |
1137 G1CollectedHeap* g1h = G1CollectedHeap::heap(); | |
1138 | |
1139 // If a full collection has happened, we shouldn't do this. | |
1140 if (has_aborted()) { | |
1141 g1h->set_marking_complete(); // So bitmap clearing isn't confused | |
1142 return; | |
1143 } | |
1144 | |
845
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
1145 if (VerifyDuringGC) { |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
1146 HandleMark hm; // handle scope |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
1147 gclog_or_tty->print(" VerifyDuringGC:(before)"); |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
1148 Universe::heap()->prepare_for_verify(); |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
1149 Universe::verify(true, false, true); |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
1150 } |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
1151 |
342 | 1152 G1CollectorPolicy* g1p = g1h->g1_policy(); |
1153 g1p->record_concurrent_mark_remark_start(); | |
1154 | |
1155 double start = os::elapsedTime(); | |
1156 | |
1157 checkpointRootsFinalWork(); | |
1158 | |
1159 double mark_work_end = os::elapsedTime(); | |
1160 | |
1161 weakRefsWork(clear_all_soft_refs); | |
1162 | |
1163 if (has_overflown()) { | |
1164 // Oops. We overflowed. Restart concurrent marking. | |
1165 _restart_for_overflow = true; | |
1166 // Clear the flag. We do not need it any more. | |
1167 clear_has_overflown(); | |
1168 if (G1TraceMarkStackOverflow) | |
1169 gclog_or_tty->print_cr("\nRemark led to restart for overflow."); | |
1170 } else { | |
1171 // We're done with marking. | |
1317
d4197f8d516a
6935821: G1: threads created during marking do not active their SATB queues
tonyp
parents:
1314
diff
changeset
|
1172 // This is the end of the marking cycle, we're expected all |
d4197f8d516a
6935821: G1: threads created during marking do not active their SATB queues
tonyp
parents:
1314
diff
changeset
|
1173 // threads to have SATB queues with active set to true. |
d4197f8d516a
6935821: G1: threads created during marking do not active their SATB queues
tonyp
parents:
1314
diff
changeset
|
1174 JavaThread::satb_mark_queue_set().set_active_all_threads( |
d4197f8d516a
6935821: G1: threads created during marking do not active their SATB queues
tonyp
parents:
1314
diff
changeset
|
1175 false, /* new active value */ |
d4197f8d516a
6935821: G1: threads created during marking do not active their SATB queues
tonyp
parents:
1314
diff
changeset
|
1176 true /* expected_active */); |
811 | 1177 |
1178 if (VerifyDuringGC) { | |
845
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
1179 HandleMark hm; // handle scope |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
1180 gclog_or_tty->print(" VerifyDuringGC:(after)"); |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
1181 Universe::heap()->prepare_for_verify(); |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
1182 Universe::heap()->verify(/* allow_dirty */ true, |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
1183 /* silent */ false, |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
1184 /* use_prev_marking */ false); |
811 | 1185 } |
342 | 1186 } |
1187 | |
1188 #if VERIFY_OBJS_PROCESSED | |
1189 _scan_obj_cl.objs_processed = 0; | |
1190 ThreadLocalObjQueue::objs_enqueued = 0; | |
1191 #endif | |
1192 | |
1193 // Statistics | |
1194 double now = os::elapsedTime(); | |
1195 _remark_mark_times.add((mark_work_end - start) * 1000.0); | |
1196 _remark_weak_ref_times.add((now - mark_work_end) * 1000.0); | |
1197 _remark_times.add((now - start) * 1000.0); | |
1198 | |
1199 g1p->record_concurrent_mark_remark_end(); | |
1200 } | |
1201 | |
1202 | |
1203 #define CARD_BM_TEST_MODE 0 | |
1204 | |
1205 class CalcLiveObjectsClosure: public HeapRegionClosure { | |
1206 | |
1207 CMBitMapRO* _bm; | |
1208 ConcurrentMark* _cm; | |
1209 bool _changed; | |
1210 bool _yield; | |
1211 size_t _words_done; | |
1212 size_t _tot_live; | |
1213 size_t _tot_used; | |
1214 size_t _regions_done; | |
1215 double _start_vtime_sec; | |
1216 | |
1217 BitMap* _region_bm; | |
1218 BitMap* _card_bm; | |
1219 intptr_t _bottom_card_num; | |
1220 bool _final; | |
1221 | |
1222 void mark_card_num_range(intptr_t start_card_num, intptr_t last_card_num) { | |
1223 for (intptr_t i = start_card_num; i <= last_card_num; i++) { | |
1224 #if CARD_BM_TEST_MODE | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
1225 guarantee(_card_bm->at(i - _bottom_card_num), "Should already be set."); |
342 | 1226 #else |
1227 _card_bm->par_at_put(i - _bottom_card_num, 1); | |
1228 #endif | |
1229 } | |
1230 } | |
1231 | |
1232 public: | |
1233 CalcLiveObjectsClosure(bool final, | |
1234 CMBitMapRO *bm, ConcurrentMark *cm, | |
936 | 1235 BitMap* region_bm, BitMap* card_bm) : |
342 | 1236 _bm(bm), _cm(cm), _changed(false), _yield(true), |
1237 _words_done(0), _tot_live(0), _tot_used(0), | |
936 | 1238 _region_bm(region_bm), _card_bm(card_bm),_final(final), |
342 | 1239 _regions_done(0), _start_vtime_sec(0.0) |
1240 { | |
1241 _bottom_card_num = | |
1242 intptr_t(uintptr_t(G1CollectedHeap::heap()->reserved_region().start()) >> | |
1243 CardTableModRefBS::card_shift); | |
1244 } | |
1245 | |
829 | 1246 // It takes a region that's not empty (i.e., it has at least one |
1247 // live object in it and sets its corresponding bit on the region | |
1248 // bitmap to 1. If the region is "starts humongous" it will also set | |
1249 // to 1 the bits on the region bitmap that correspond to its | |
1250 // associated "continues humongous" regions. | |
1251 void set_bit_for_region(HeapRegion* hr) { | |
1252 assert(!hr->continuesHumongous(), "should have filtered those out"); | |
1253 | |
1254 size_t index = hr->hrs_index(); | |
1255 if (!hr->startsHumongous()) { | |
1256 // Normal (non-humongous) case: just set the bit. | |
1257 _region_bm->par_at_put((BitMap::idx_t) index, true); | |
1258 } else { | |
1259 // Starts humongous case: calculate how many regions are part of | |
1260 // this humongous region and then set the bit range. It might | |
1261 // have been a bit more efficient to look at the object that | |
1262 // spans these humongous regions to calculate their number from | |
1263 // the object's size. However, it's a good idea to calculate | |
1264 // this based on the metadata itself, and not the region | |
1265 // contents, so that this code is not aware of what goes into | |
1266 // the humongous regions (in case this changes in the future). | |
1267 G1CollectedHeap* g1h = G1CollectedHeap::heap(); | |
1268 size_t end_index = index + 1; | |
831 | 1269 while (end_index < g1h->n_regions()) { |
1270 HeapRegion* chr = g1h->region_at(end_index); | |
829 | 1271 if (!chr->continuesHumongous()) { |
1272 break; | |
1273 } | |
1274 end_index += 1; | |
1275 } | |
1276 _region_bm->par_at_put_range((BitMap::idx_t) index, | |
1277 (BitMap::idx_t) end_index, true); | |
1278 } | |
1279 } | |
1280 | |
342 | 1281 bool doHeapRegion(HeapRegion* hr) { |
1282 if (!_final && _regions_done == 0) | |
1283 _start_vtime_sec = os::elapsedVTime(); | |
1284 | |
639 | 1285 if (hr->continuesHumongous()) { |
829 | 1286 // We will ignore these here and process them when their |
1287 // associated "starts humongous" region is processed (see | |
1288 // set_bit_for_heap_region()). Note that we cannot rely on their | |
1289 // associated "starts humongous" region to have their bit set to | |
1290 // 1 since, due to the region chunking in the parallel region | |
1291 // iteration, a "continues humongous" region might be visited | |
1292 // before its associated "starts humongous". | |
639 | 1293 return false; |
1294 } | |
342 | 1295 |
1296 HeapWord* nextTop = hr->next_top_at_mark_start(); | |
1297 HeapWord* start = hr->top_at_conc_mark_count(); | |
1298 assert(hr->bottom() <= start && start <= hr->end() && | |
1299 hr->bottom() <= nextTop && nextTop <= hr->end() && | |
1300 start <= nextTop, | |
1301 "Preconditions."); | |
1302 // Otherwise, record the number of word's we'll examine. | |
1303 size_t words_done = (nextTop - start); | |
1304 // Find the first marked object at or after "start". | |
1305 start = _bm->getNextMarkedWordAddress(start, nextTop); | |
1306 size_t marked_bytes = 0; | |
1307 | |
1308 // Below, the term "card num" means the result of shifting an address | |
1309 // by the card shift -- address 0 corresponds to card number 0. One | |
1310 // must subtract the card num of the bottom of the heap to obtain a | |
1311 // card table index. | |
1312 // The first card num of the sequence of live cards currently being | |
1313 // constructed. -1 ==> no sequence. | |
1314 intptr_t start_card_num = -1; | |
1315 // The last card num of the sequence of live cards currently being | |
1316 // constructed. -1 ==> no sequence. | |
1317 intptr_t last_card_num = -1; | |
1318 | |
1319 while (start < nextTop) { | |
1320 if (_yield && _cm->do_yield_check()) { | |
1321 // We yielded. It might be for a full collection, in which case | |
1322 // all bets are off; terminate the traversal. | |
1323 if (_cm->has_aborted()) { | |
1324 _changed = false; | |
1325 return true; | |
1326 } else { | |
1327 // Otherwise, it might be a collection pause, and the region | |
1328 // we're looking at might be in the collection set. We'll | |
1329 // abandon this region. | |
1330 return false; | |
1331 } | |
1332 } | |
1333 oop obj = oop(start); | |
1334 int obj_sz = obj->size(); | |
1335 // The card num of the start of the current object. | |
1336 intptr_t obj_card_num = | |
1337 intptr_t(uintptr_t(start) >> CardTableModRefBS::card_shift); | |
1338 | |
1339 HeapWord* obj_last = start + obj_sz - 1; | |
1340 intptr_t obj_last_card_num = | |
1341 intptr_t(uintptr_t(obj_last) >> CardTableModRefBS::card_shift); | |
1342 | |
1343 if (obj_card_num != last_card_num) { | |
1344 if (start_card_num == -1) { | |
1345 assert(last_card_num == -1, "Both or neither."); | |
1346 start_card_num = obj_card_num; | |
1347 } else { | |
1348 assert(last_card_num != -1, "Both or neither."); | |
1349 assert(obj_card_num >= last_card_num, "Inv"); | |
1350 if ((obj_card_num - last_card_num) > 1) { | |
1351 // Mark the last run, and start a new one. | |
1352 mark_card_num_range(start_card_num, last_card_num); | |
1353 start_card_num = obj_card_num; | |
1354 } | |
1355 } | |
1356 #if CARD_BM_TEST_MODE | |
1357 /* | |
1358 gclog_or_tty->print_cr("Setting bits from %d/%d.", | |
1359 obj_card_num - _bottom_card_num, | |
1360 obj_last_card_num - _bottom_card_num); | |
1361 */ | |
1362 for (intptr_t j = obj_card_num; j <= obj_last_card_num; j++) { | |
1363 _card_bm->par_at_put(j - _bottom_card_num, 1); | |
1364 } | |
1365 #endif | |
1366 } | |
1367 // In any case, we set the last card num. | |
1368 last_card_num = obj_last_card_num; | |
1369 | |
1030
dfdaf65c3423
6858886: G1: guarantee(_next_marked_bytes <= used(),"invariant") at heapRegion.hpp:359
apetrusenko
parents:
1023
diff
changeset
|
1370 marked_bytes += (size_t)obj_sz * HeapWordSize; |
342 | 1371 // Find the next marked object after this one. |
1372 start = _bm->getNextMarkedWordAddress(start + 1, nextTop); | |
1373 _changed = true; | |
1374 } | |
1375 // Handle the last range, if any. | |
1376 if (start_card_num != -1) | |
1377 mark_card_num_range(start_card_num, last_card_num); | |
1378 if (_final) { | |
1379 // Mark the allocated-since-marking portion... | |
1380 HeapWord* tp = hr->top(); | |
1381 if (nextTop < tp) { | |
1382 start_card_num = | |
1383 intptr_t(uintptr_t(nextTop) >> CardTableModRefBS::card_shift); | |
1384 last_card_num = | |
1385 intptr_t(uintptr_t(tp) >> CardTableModRefBS::card_shift); | |
1386 mark_card_num_range(start_card_num, last_card_num); | |
1387 // This definitely means the region has live objects. | |
829 | 1388 set_bit_for_region(hr); |
342 | 1389 } |
1390 } | |
1391 | |
1392 hr->add_to_marked_bytes(marked_bytes); | |
1393 // Update the live region bitmap. | |
1394 if (marked_bytes > 0) { | |
829 | 1395 set_bit_for_region(hr); |
342 | 1396 } |
1397 hr->set_top_at_conc_mark_count(nextTop); | |
1398 _tot_live += hr->next_live_bytes(); | |
1399 _tot_used += hr->used(); | |
1400 _words_done = words_done; | |
1401 | |
1402 if (!_final) { | |
1403 ++_regions_done; | |
1404 if (_regions_done % 10 == 0) { | |
1405 double end_vtime_sec = os::elapsedVTime(); | |
1406 double elapsed_vtime_sec = end_vtime_sec - _start_vtime_sec; | |
1407 if (elapsed_vtime_sec > (10.0 / 1000.0)) { | |
1408 jlong sleep_time_ms = | |
1409 (jlong) (elapsed_vtime_sec * _cm->cleanup_sleep_factor() * 1000.0); | |
1410 os::sleep(Thread::current(), sleep_time_ms, false); | |
1411 _start_vtime_sec = end_vtime_sec; | |
1412 } | |
1413 } | |
1414 } | |
1415 | |
1416 return false; | |
1417 } | |
1418 | |
1419 bool changed() { return _changed; } | |
1420 void reset() { _changed = false; _words_done = 0; } | |
1421 void no_yield() { _yield = false; } | |
1422 size_t words_done() { return _words_done; } | |
1423 size_t tot_live() { return _tot_live; } | |
1424 size_t tot_used() { return _tot_used; } | |
1425 }; | |
1426 | |
1427 | |
1428 void ConcurrentMark::calcDesiredRegions() { | |
1429 _region_bm.clear(); | |
1430 _card_bm.clear(); | |
1431 CalcLiveObjectsClosure calccl(false /*final*/, | |
1432 nextMarkBitMap(), this, | |
936 | 1433 &_region_bm, &_card_bm); |
342 | 1434 G1CollectedHeap *g1h = G1CollectedHeap::heap(); |
1435 g1h->heap_region_iterate(&calccl); | |
1436 | |
1437 do { | |
1438 calccl.reset(); | |
1439 g1h->heap_region_iterate(&calccl); | |
1440 } while (calccl.changed()); | |
1441 } | |
1442 | |
1443 class G1ParFinalCountTask: public AbstractGangTask { | |
1444 protected: | |
1445 G1CollectedHeap* _g1h; | |
1446 CMBitMap* _bm; | |
1447 size_t _n_workers; | |
1448 size_t *_live_bytes; | |
1449 size_t *_used_bytes; | |
1450 BitMap* _region_bm; | |
1451 BitMap* _card_bm; | |
1452 public: | |
1453 G1ParFinalCountTask(G1CollectedHeap* g1h, CMBitMap* bm, | |
1454 BitMap* region_bm, BitMap* card_bm) : | |
1455 AbstractGangTask("G1 final counting"), _g1h(g1h), | |
1456 _bm(bm), _region_bm(region_bm), _card_bm(card_bm) | |
1457 { | |
1458 if (ParallelGCThreads > 0) | |
1459 _n_workers = _g1h->workers()->total_workers(); | |
1460 else | |
1461 _n_workers = 1; | |
1462 _live_bytes = NEW_C_HEAP_ARRAY(size_t, _n_workers); | |
1463 _used_bytes = NEW_C_HEAP_ARRAY(size_t, _n_workers); | |
1464 } | |
1465 | |
1466 ~G1ParFinalCountTask() { | |
1467 FREE_C_HEAP_ARRAY(size_t, _live_bytes); | |
1468 FREE_C_HEAP_ARRAY(size_t, _used_bytes); | |
1469 } | |
1470 | |
1471 void work(int i) { | |
1472 CalcLiveObjectsClosure calccl(true /*final*/, | |
1473 _bm, _g1h->concurrent_mark(), | |
936 | 1474 _region_bm, _card_bm); |
342 | 1475 calccl.no_yield(); |
1833
8b10f48633dc
6984287: Regularize how GC parallel workers are specified.
jmasa
parents:
1719
diff
changeset
|
1476 if (G1CollectedHeap::use_parallel_gc_threads()) { |
355 | 1477 _g1h->heap_region_par_iterate_chunked(&calccl, i, |
1478 HeapRegion::FinalCountClaimValue); | |
342 | 1479 } else { |
1480 _g1h->heap_region_iterate(&calccl); | |
1481 } | |
1482 assert(calccl.complete(), "Shouldn't have yielded!"); | |
1483 | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
1484 assert((size_t) i < _n_workers, "invariant"); |
342 | 1485 _live_bytes[i] = calccl.tot_live(); |
1486 _used_bytes[i] = calccl.tot_used(); | |
1487 } | |
1488 size_t live_bytes() { | |
1489 size_t live_bytes = 0; | |
1490 for (size_t i = 0; i < _n_workers; ++i) | |
1491 live_bytes += _live_bytes[i]; | |
1492 return live_bytes; | |
1493 } | |
1494 size_t used_bytes() { | |
1495 size_t used_bytes = 0; | |
1496 for (size_t i = 0; i < _n_workers; ++i) | |
1497 used_bytes += _used_bytes[i]; | |
1498 return used_bytes; | |
1499 } | |
1500 }; | |
1501 | |
1502 class G1ParNoteEndTask; | |
1503 | |
1504 class G1NoteEndOfConcMarkClosure : public HeapRegionClosure { | |
1505 G1CollectedHeap* _g1; | |
1506 int _worker_num; | |
1507 size_t _max_live_bytes; | |
1508 size_t _regions_claimed; | |
1509 size_t _freed_bytes; | |
1510 size_t _cleared_h_regions; | |
1511 size_t _freed_regions; | |
1512 UncleanRegionList* _unclean_region_list; | |
1513 double _claimed_region_time; | |
1514 double _max_region_time; | |
1515 | |
1516 public: | |
1517 G1NoteEndOfConcMarkClosure(G1CollectedHeap* g1, | |
1518 UncleanRegionList* list, | |
1519 int worker_num); | |
1520 size_t freed_bytes() { return _freed_bytes; } | |
1521 size_t cleared_h_regions() { return _cleared_h_regions; } | |
1522 size_t freed_regions() { return _freed_regions; } | |
1523 UncleanRegionList* unclean_region_list() { | |
1524 return _unclean_region_list; | |
1525 } | |
1526 | |
1527 bool doHeapRegion(HeapRegion *r); | |
1528 | |
1529 size_t max_live_bytes() { return _max_live_bytes; } | |
1530 size_t regions_claimed() { return _regions_claimed; } | |
1531 double claimed_region_time_sec() { return _claimed_region_time; } | |
1532 double max_region_time_sec() { return _max_region_time; } | |
1533 }; | |
1534 | |
1535 class G1ParNoteEndTask: public AbstractGangTask { | |
1536 friend class G1NoteEndOfConcMarkClosure; | |
1537 protected: | |
1538 G1CollectedHeap* _g1h; | |
1539 size_t _max_live_bytes; | |
1540 size_t _freed_bytes; | |
1541 ConcurrentMark::ParCleanupThreadState** _par_cleanup_thread_state; | |
1542 public: | |
1543 G1ParNoteEndTask(G1CollectedHeap* g1h, | |
1544 ConcurrentMark::ParCleanupThreadState** | |
1545 par_cleanup_thread_state) : | |
1546 AbstractGangTask("G1 note end"), _g1h(g1h), | |
1547 _max_live_bytes(0), _freed_bytes(0), | |
1548 _par_cleanup_thread_state(par_cleanup_thread_state) | |
1549 {} | |
1550 | |
1551 void work(int i) { | |
1552 double start = os::elapsedTime(); | |
1553 G1NoteEndOfConcMarkClosure g1_note_end(_g1h, | |
1554 &_par_cleanup_thread_state[i]->list, | |
1555 i); | |
1833
8b10f48633dc
6984287: Regularize how GC parallel workers are specified.
jmasa
parents:
1719
diff
changeset
|
1556 if (G1CollectedHeap::use_parallel_gc_threads()) { |
355 | 1557 _g1h->heap_region_par_iterate_chunked(&g1_note_end, i, |
1558 HeapRegion::NoteEndClaimValue); | |
342 | 1559 } else { |
1560 _g1h->heap_region_iterate(&g1_note_end); | |
1561 } | |
1562 assert(g1_note_end.complete(), "Shouldn't have yielded!"); | |
1563 | |
1564 // Now finish up freeing the current thread's regions. | |
1565 _g1h->finish_free_region_work(g1_note_end.freed_bytes(), | |
1566 g1_note_end.cleared_h_regions(), | |
1567 0, NULL); | |
1568 { | |
1569 MutexLockerEx x(ParGCRareEvent_lock, Mutex::_no_safepoint_check_flag); | |
1570 _max_live_bytes += g1_note_end.max_live_bytes(); | |
1571 _freed_bytes += g1_note_end.freed_bytes(); | |
1572 } | |
1573 double end = os::elapsedTime(); | |
1574 if (G1PrintParCleanupStats) { | |
1575 gclog_or_tty->print(" Worker thread %d [%8.3f..%8.3f = %8.3f ms] " | |
1576 "claimed %d regions (tot = %8.3f ms, max = %8.3f ms).\n", | |
1577 i, start, end, (end-start)*1000.0, | |
1578 g1_note_end.regions_claimed(), | |
1579 g1_note_end.claimed_region_time_sec()*1000.0, | |
1580 g1_note_end.max_region_time_sec()*1000.0); | |
1581 } | |
1582 } | |
1583 size_t max_live_bytes() { return _max_live_bytes; } | |
1584 size_t freed_bytes() { return _freed_bytes; } | |
1585 }; | |
1586 | |
1587 class G1ParScrubRemSetTask: public AbstractGangTask { | |
1588 protected: | |
1589 G1RemSet* _g1rs; | |
1590 BitMap* _region_bm; | |
1591 BitMap* _card_bm; | |
1592 public: | |
1593 G1ParScrubRemSetTask(G1CollectedHeap* g1h, | |
1594 BitMap* region_bm, BitMap* card_bm) : | |
1595 AbstractGangTask("G1 ScrubRS"), _g1rs(g1h->g1_rem_set()), | |
1596 _region_bm(region_bm), _card_bm(card_bm) | |
1597 {} | |
1598 | |
1599 void work(int i) { | |
1833
8b10f48633dc
6984287: Regularize how GC parallel workers are specified.
jmasa
parents:
1719
diff
changeset
|
1600 if (G1CollectedHeap::use_parallel_gc_threads()) { |
355 | 1601 _g1rs->scrub_par(_region_bm, _card_bm, i, |
1602 HeapRegion::ScrubRemSetClaimValue); | |
342 | 1603 } else { |
1604 _g1rs->scrub(_region_bm, _card_bm); | |
1605 } | |
1606 } | |
1607 | |
1608 }; | |
1609 | |
1610 G1NoteEndOfConcMarkClosure:: | |
1611 G1NoteEndOfConcMarkClosure(G1CollectedHeap* g1, | |
1612 UncleanRegionList* list, | |
1613 int worker_num) | |
1614 : _g1(g1), _worker_num(worker_num), | |
1615 _max_live_bytes(0), _regions_claimed(0), | |
1616 _freed_bytes(0), _cleared_h_regions(0), _freed_regions(0), | |
1617 _claimed_region_time(0.0), _max_region_time(0.0), | |
1618 _unclean_region_list(list) | |
1619 {} | |
1620 | |
1621 bool G1NoteEndOfConcMarkClosure::doHeapRegion(HeapRegion *r) { | |
1622 // We use a claim value of zero here because all regions | |
1623 // were claimed with value 1 in the FinalCount task. | |
1624 r->reset_gc_time_stamp(); | |
1625 if (!r->continuesHumongous()) { | |
1626 double start = os::elapsedTime(); | |
1627 _regions_claimed++; | |
1628 r->note_end_of_marking(); | |
1629 _max_live_bytes += r->max_live_bytes(); | |
1630 _g1->free_region_if_totally_empty_work(r, | |
1631 _freed_bytes, | |
1632 _cleared_h_regions, | |
1633 _freed_regions, | |
1634 _unclean_region_list, | |
1635 true /*par*/); | |
1636 double region_time = (os::elapsedTime() - start); | |
1637 _claimed_region_time += region_time; | |
1638 if (region_time > _max_region_time) _max_region_time = region_time; | |
1639 } | |
1640 return false; | |
1641 } | |
1642 | |
1643 void ConcurrentMark::cleanup() { | |
1644 // world is stopped at this checkpoint | |
1645 assert(SafepointSynchronize::is_at_safepoint(), | |
1646 "world should be stopped"); | |
1647 G1CollectedHeap* g1h = G1CollectedHeap::heap(); | |
1648 | |
1649 // If a full collection has happened, we shouldn't do this. | |
1650 if (has_aborted()) { | |
1651 g1h->set_marking_complete(); // So bitmap clearing isn't confused | |
1652 return; | |
1653 } | |
1654 | |
845
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
1655 if (VerifyDuringGC) { |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
1656 HandleMark hm; // handle scope |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
1657 gclog_or_tty->print(" VerifyDuringGC:(before)"); |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
1658 Universe::heap()->prepare_for_verify(); |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
1659 Universe::verify(/* allow dirty */ true, |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
1660 /* silent */ false, |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
1661 /* prev marking */ true); |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
1662 } |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
1663 |
342 | 1664 G1CollectorPolicy* g1p = G1CollectedHeap::heap()->g1_policy(); |
1665 g1p->record_concurrent_mark_cleanup_start(); | |
1666 | |
1667 double start = os::elapsedTime(); | |
1668 | |
1669 // Do counting once more with the world stopped for good measure. | |
1670 G1ParFinalCountTask g1_par_count_task(g1h, nextMarkBitMap(), | |
1671 &_region_bm, &_card_bm); | |
1833
8b10f48633dc
6984287: Regularize how GC parallel workers are specified.
jmasa
parents:
1719
diff
changeset
|
1672 if (G1CollectedHeap::use_parallel_gc_threads()) { |
355 | 1673 assert(g1h->check_heap_region_claim_values( |
1674 HeapRegion::InitialClaimValue), | |
1675 "sanity check"); | |
1676 | |
342 | 1677 int n_workers = g1h->workers()->total_workers(); |
1678 g1h->set_par_threads(n_workers); | |
1679 g1h->workers()->run_task(&g1_par_count_task); | |
1680 g1h->set_par_threads(0); | |
355 | 1681 |
1682 assert(g1h->check_heap_region_claim_values( | |
1683 HeapRegion::FinalCountClaimValue), | |
1684 "sanity check"); | |
342 | 1685 } else { |
1686 g1_par_count_task.work(0); | |
1687 } | |
1688 | |
1689 size_t known_garbage_bytes = | |
1690 g1_par_count_task.used_bytes() - g1_par_count_task.live_bytes(); | |
1691 #if 0 | |
1692 gclog_or_tty->print_cr("used %1.2lf, live %1.2lf, garbage %1.2lf", | |
1693 (double) g1_par_count_task.used_bytes() / (double) (1024 * 1024), | |
1694 (double) g1_par_count_task.live_bytes() / (double) (1024 * 1024), | |
1695 (double) known_garbage_bytes / (double) (1024 * 1024)); | |
1696 #endif // 0 | |
1697 g1p->set_known_garbage_bytes(known_garbage_bytes); | |
1698 | |
1699 size_t start_used_bytes = g1h->used(); | |
1700 _at_least_one_mark_complete = true; | |
1701 g1h->set_marking_complete(); | |
1702 | |
1703 double count_end = os::elapsedTime(); | |
1704 double this_final_counting_time = (count_end - start); | |
1705 if (G1PrintParCleanupStats) { | |
1706 gclog_or_tty->print_cr("Cleanup:"); | |
1707 gclog_or_tty->print_cr(" Finalize counting: %8.3f ms", | |
1708 this_final_counting_time*1000.0); | |
1709 } | |
1710 _total_counting_time += this_final_counting_time; | |
1711 | |
1712 // Install newly created mark bitMap as "prev". | |
1713 swapMarkBitMaps(); | |
1714 | |
1715 g1h->reset_gc_time_stamp(); | |
1716 | |
1717 // Note end of marking in all heap regions. | |
1718 double note_end_start = os::elapsedTime(); | |
1719 G1ParNoteEndTask g1_par_note_end_task(g1h, _par_cleanup_thread_state); | |
1833
8b10f48633dc
6984287: Regularize how GC parallel workers are specified.
jmasa
parents:
1719
diff
changeset
|
1720 if (G1CollectedHeap::use_parallel_gc_threads()) { |
342 | 1721 int n_workers = g1h->workers()->total_workers(); |
1722 g1h->set_par_threads(n_workers); | |
1723 g1h->workers()->run_task(&g1_par_note_end_task); | |
1724 g1h->set_par_threads(0); | |
355 | 1725 |
1726 assert(g1h->check_heap_region_claim_values(HeapRegion::NoteEndClaimValue), | |
1727 "sanity check"); | |
342 | 1728 } else { |
1729 g1_par_note_end_task.work(0); | |
1730 } | |
1731 g1h->set_unclean_regions_coming(true); | |
1732 double note_end_end = os::elapsedTime(); | |
1733 // Tell the mutators that there might be unclean regions coming... | |
1734 if (G1PrintParCleanupStats) { | |
1735 gclog_or_tty->print_cr(" note end of marking: %8.3f ms.", | |
1736 (note_end_end - note_end_start)*1000.0); | |
1737 } | |
1738 | |
355 | 1739 |
342 | 1740 // call below, since it affects the metric by which we sort the heap |
1741 // regions. | |
1742 if (G1ScrubRemSets) { | |
1743 double rs_scrub_start = os::elapsedTime(); | |
1744 G1ParScrubRemSetTask g1_par_scrub_rs_task(g1h, &_region_bm, &_card_bm); | |
1833
8b10f48633dc
6984287: Regularize how GC parallel workers are specified.
jmasa
parents:
1719
diff
changeset
|
1745 if (G1CollectedHeap::use_parallel_gc_threads()) { |
342 | 1746 int n_workers = g1h->workers()->total_workers(); |
1747 g1h->set_par_threads(n_workers); | |
1748 g1h->workers()->run_task(&g1_par_scrub_rs_task); | |
1749 g1h->set_par_threads(0); | |
355 | 1750 |
1751 assert(g1h->check_heap_region_claim_values( | |
1752 HeapRegion::ScrubRemSetClaimValue), | |
1753 "sanity check"); | |
342 | 1754 } else { |
1755 g1_par_scrub_rs_task.work(0); | |
1756 } | |
1757 | |
1758 double rs_scrub_end = os::elapsedTime(); | |
1759 double this_rs_scrub_time = (rs_scrub_end - rs_scrub_start); | |
1760 _total_rs_scrub_time += this_rs_scrub_time; | |
1761 } | |
1762 | |
1763 // this will also free any regions totally full of garbage objects, | |
1764 // and sort the regions. | |
1765 g1h->g1_policy()->record_concurrent_mark_cleanup_end( | |
1766 g1_par_note_end_task.freed_bytes(), | |
1767 g1_par_note_end_task.max_live_bytes()); | |
1768 | |
1769 // Statistics. | |
1770 double end = os::elapsedTime(); | |
1771 _cleanup_times.add((end - start) * 1000.0); | |
1772 | |
1773 // G1CollectedHeap::heap()->print(); | |
1774 // gclog_or_tty->print_cr("HEAP GC TIME STAMP : %d", | |
1775 // G1CollectedHeap::heap()->get_gc_time_stamp()); | |
1776 | |
1777 if (PrintGC || PrintGCDetails) { | |
1778 g1h->print_size_transition(gclog_or_tty, | |
1779 start_used_bytes, | |
1780 g1h->used(), | |
1781 g1h->capacity()); | |
1782 } | |
1783 | |
1784 size_t cleaned_up_bytes = start_used_bytes - g1h->used(); | |
1785 g1p->decrease_known_garbage_bytes(cleaned_up_bytes); | |
1786 | |
1787 // We need to make this be a "collection" so any collection pause that | |
1788 // races with it goes around and waits for completeCleanup to finish. | |
1789 g1h->increment_total_collections(); | |
1790 | |
751 | 1791 if (VerifyDuringGC) { |
845
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
1792 HandleMark hm; // handle scope |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
1793 gclog_or_tty->print(" VerifyDuringGC:(after)"); |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
1794 Universe::heap()->prepare_for_verify(); |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
1795 Universe::verify(/* allow dirty */ true, |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
1796 /* silent */ false, |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
1797 /* prev marking */ true); |
342 | 1798 } |
1799 } | |
1800 | |
1801 void ConcurrentMark::completeCleanup() { | |
1802 // A full collection intervened. | |
1803 if (has_aborted()) return; | |
1804 | |
1805 int first = 0; | |
1806 int last = (int)MAX2(ParallelGCThreads, (size_t)1); | |
1807 for (int t = 0; t < last; t++) { | |
1808 UncleanRegionList* list = &_par_cleanup_thread_state[t]->list; | |
1809 assert(list->well_formed(), "Inv"); | |
1810 HeapRegion* hd = list->hd(); | |
1811 while (hd != NULL) { | |
1812 // Now finish up the other stuff. | |
1813 hd->rem_set()->clear(); | |
1814 HeapRegion* next_hd = hd->next_from_unclean_list(); | |
1815 (void)list->pop(); | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
1816 assert(list->hd() == next_hd, "how not?"); |
342 | 1817 _g1h->put_region_on_unclean_list(hd); |
1818 if (!hd->isHumongous()) { | |
1819 // Add this to the _free_regions count by 1. | |
1820 _g1h->finish_free_region_work(0, 0, 1, NULL); | |
1821 } | |
1822 hd = list->hd(); | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
1823 assert(hd == next_hd, "how not?"); |
342 | 1824 } |
1825 } | |
1826 } | |
1827 | |
1828 | |
1829 class G1CMIsAliveClosure: public BoolObjectClosure { | |
1830 G1CollectedHeap* _g1; | |
1831 public: | |
1832 G1CMIsAliveClosure(G1CollectedHeap* g1) : | |
1833 _g1(g1) | |
1834 {} | |
1835 | |
1836 void do_object(oop obj) { | |
1837 assert(false, "not to be invoked"); | |
1838 } | |
1839 bool do_object_b(oop obj) { | |
1840 HeapWord* addr = (HeapWord*)obj; | |
1841 return addr != NULL && | |
1842 (!_g1->is_in_g1_reserved(addr) || !_g1->is_obj_ill(obj)); | |
1843 } | |
1844 }; | |
1845 | |
1846 class G1CMKeepAliveClosure: public OopClosure { | |
1847 G1CollectedHeap* _g1; | |
1848 ConcurrentMark* _cm; | |
1849 CMBitMap* _bitMap; | |
1850 public: | |
1851 G1CMKeepAliveClosure(G1CollectedHeap* g1, ConcurrentMark* cm, | |
1852 CMBitMap* bitMap) : | |
1853 _g1(g1), _cm(cm), | |
1854 _bitMap(bitMap) {} | |
1855 | |
845
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
1856 virtual void do_oop(narrowOop* p) { do_oop_work(p); } |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
1857 virtual void do_oop( oop* p) { do_oop_work(p); } |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
1858 |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
1859 template <class T> void do_oop_work(T* p) { |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
1860 oop thisOop = oopDesc::load_decode_heap_oop(p); |
342 | 1861 HeapWord* addr = (HeapWord*)thisOop; |
1862 if (_g1->is_in_g1_reserved(addr) && _g1->is_obj_ill(thisOop)) { | |
1863 _bitMap->mark(addr); | |
1864 _cm->mark_stack_push(thisOop); | |
1865 } | |
1866 } | |
1867 }; | |
1868 | |
1869 class G1CMDrainMarkingStackClosure: public VoidClosure { | |
1870 CMMarkStack* _markStack; | |
1871 CMBitMap* _bitMap; | |
1872 G1CMKeepAliveClosure* _oopClosure; | |
1873 public: | |
1874 G1CMDrainMarkingStackClosure(CMBitMap* bitMap, CMMarkStack* markStack, | |
1875 G1CMKeepAliveClosure* oopClosure) : | |
1876 _bitMap(bitMap), | |
1877 _markStack(markStack), | |
1878 _oopClosure(oopClosure) | |
1879 {} | |
1880 | |
1881 void do_void() { | |
1882 _markStack->drain((OopClosure*)_oopClosure, _bitMap, false); | |
1883 } | |
1884 }; | |
1885 | |
1886 void ConcurrentMark::weakRefsWork(bool clear_all_soft_refs) { | |
1887 ResourceMark rm; | |
1888 HandleMark hm; | |
453
c96030fff130
6684579: SoftReference processing can be made more efficient
ysr
parents:
355
diff
changeset
|
1889 G1CollectedHeap* g1h = G1CollectedHeap::heap(); |
c96030fff130
6684579: SoftReference processing can be made more efficient
ysr
parents:
355
diff
changeset
|
1890 ReferenceProcessor* rp = g1h->ref_processor(); |
342 | 1891 |
1974
fd1d227ef1b9
6983204: G1: Nightly test nsk/regression/b4958615 failing with +ExplicitGCInvokesConcurrent
johnc
parents:
1972
diff
changeset
|
1892 // See the comment in G1CollectedHeap::ref_processing_init() |
fd1d227ef1b9
6983204: G1: Nightly test nsk/regression/b4958615 failing with +ExplicitGCInvokesConcurrent
johnc
parents:
1972
diff
changeset
|
1893 // about how reference processing currently works in G1. |
fd1d227ef1b9
6983204: G1: Nightly test nsk/regression/b4958615 failing with +ExplicitGCInvokesConcurrent
johnc
parents:
1972
diff
changeset
|
1894 |
342 | 1895 // Process weak references. |
457
27a80744a83b
6778647: snap(), snap_policy() should be renamed setup(), setup_policy()
ysr
parents:
453
diff
changeset
|
1896 rp->setup_policy(clear_all_soft_refs); |
342 | 1897 assert(_markStack.isEmpty(), "mark stack should be empty"); |
1898 | |
453
c96030fff130
6684579: SoftReference processing can be made more efficient
ysr
parents:
355
diff
changeset
|
1899 G1CMIsAliveClosure g1IsAliveClosure (g1h); |
c96030fff130
6684579: SoftReference processing can be made more efficient
ysr
parents:
355
diff
changeset
|
1900 G1CMKeepAliveClosure g1KeepAliveClosure(g1h, this, nextMarkBitMap()); |
342 | 1901 G1CMDrainMarkingStackClosure |
1902 g1DrainMarkingStackClosure(nextMarkBitMap(), &_markStack, | |
1903 &g1KeepAliveClosure); | |
1904 | |
1905 // XXXYYY Also: copy the parallel ref processing code from CMS. | |
453
c96030fff130
6684579: SoftReference processing can be made more efficient
ysr
parents:
355
diff
changeset
|
1906 rp->process_discovered_references(&g1IsAliveClosure, |
342 | 1907 &g1KeepAliveClosure, |
1908 &g1DrainMarkingStackClosure, | |
1909 NULL); | |
1910 assert(_markStack.overflow() || _markStack.isEmpty(), | |
1911 "mark stack should be empty (unless it overflowed)"); | |
1912 if (_markStack.overflow()) { | |
1913 set_has_overflown(); | |
1914 } | |
1915 | |
1916 rp->enqueue_discovered_references(); | |
1917 rp->verify_no_references_recorded(); | |
1918 assert(!rp->discovery_enabled(), "should have been disabled"); | |
1919 | |
1920 // Now clean up stale oops in SymbolTable and StringTable | |
1921 SymbolTable::unlink(&g1IsAliveClosure); | |
1922 StringTable::unlink(&g1IsAliveClosure); | |
1923 } | |
1924 | |
1925 void ConcurrentMark::swapMarkBitMaps() { | |
1926 CMBitMapRO* temp = _prevMarkBitMap; | |
1927 _prevMarkBitMap = (CMBitMapRO*)_nextMarkBitMap; | |
1928 _nextMarkBitMap = (CMBitMap*) temp; | |
1929 } | |
1930 | |
1931 class CMRemarkTask: public AbstractGangTask { | |
1932 private: | |
1933 ConcurrentMark *_cm; | |
1934 | |
1935 public: | |
1936 void work(int worker_i) { | |
1937 // Since all available tasks are actually started, we should | |
1938 // only proceed if we're supposed to be actived. | |
1939 if ((size_t)worker_i < _cm->active_tasks()) { | |
1940 CMTask* task = _cm->task(worker_i); | |
1941 task->record_start_time(); | |
1942 do { | |
1943 task->do_marking_step(1000000000.0 /* something very large */); | |
1944 } while (task->has_aborted() && !_cm->has_overflown()); | |
1945 // If we overflow, then we do not want to restart. We instead | |
1946 // want to abort remark and do concurrent marking again. | |
1947 task->record_end_time(); | |
1948 } | |
1949 } | |
1950 | |
1951 CMRemarkTask(ConcurrentMark* cm) : | |
1952 AbstractGangTask("Par Remark"), _cm(cm) { } | |
1953 }; | |
1954 | |
1955 void ConcurrentMark::checkpointRootsFinalWork() { | |
1956 ResourceMark rm; | |
1957 HandleMark hm; | |
1958 G1CollectedHeap* g1h = G1CollectedHeap::heap(); | |
1959 | |
1960 g1h->ensure_parsability(false); | |
1961 | |
1833
8b10f48633dc
6984287: Regularize how GC parallel workers are specified.
jmasa
parents:
1719
diff
changeset
|
1962 if (G1CollectedHeap::use_parallel_gc_threads()) { |
989
148e5441d916
6863023: need non-perm oops in code cache for JSR 292
jrose
parents:
912
diff
changeset
|
1963 G1CollectedHeap::StrongRootsScope srs(g1h); |
342 | 1964 // this is remark, so we'll use up all available threads |
1965 int active_workers = ParallelGCThreads; | |
1966 set_phase(active_workers, false); | |
1967 | |
1968 CMRemarkTask remarkTask(this); | |
1969 // We will start all available threads, even if we decide that the | |
1970 // active_workers will be fewer. The extra ones will just bail out | |
1971 // immediately. | |
1972 int n_workers = g1h->workers()->total_workers(); | |
1973 g1h->set_par_threads(n_workers); | |
1974 g1h->workers()->run_task(&remarkTask); | |
1975 g1h->set_par_threads(0); | |
1976 } else { | |
989
148e5441d916
6863023: need non-perm oops in code cache for JSR 292
jrose
parents:
912
diff
changeset
|
1977 G1CollectedHeap::StrongRootsScope srs(g1h); |
342 | 1978 // this is remark, so we'll use up all available threads |
1979 int active_workers = 1; | |
1980 set_phase(active_workers, false); | |
1981 | |
1982 CMRemarkTask remarkTask(this); | |
1983 // We will start all available threads, even if we decide that the | |
1984 // active_workers will be fewer. The extra ones will just bail out | |
1985 // immediately. | |
1986 remarkTask.work(0); | |
1987 } | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
1988 SATBMarkQueueSet& satb_mq_set = JavaThread::satb_mark_queue_set(); |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
1989 guarantee(satb_mq_set.completed_buffers_num() == 0, "invariant"); |
342 | 1990 |
1991 print_stats(); | |
1992 | |
1993 if (!restart_for_overflow()) | |
1994 set_non_marking_state(); | |
1995 | |
1996 #if VERIFY_OBJS_PROCESSED | |
1997 if (_scan_obj_cl.objs_processed != ThreadLocalObjQueue::objs_enqueued) { | |
1998 gclog_or_tty->print_cr("Processed = %d, enqueued = %d.", | |
1999 _scan_obj_cl.objs_processed, | |
2000 ThreadLocalObjQueue::objs_enqueued); | |
2001 guarantee(_scan_obj_cl.objs_processed == | |
2002 ThreadLocalObjQueue::objs_enqueued, | |
2003 "Different number of objs processed and enqueued."); | |
2004 } | |
2005 #endif | |
2006 } | |
2007 | |
1044 | 2008 #ifndef PRODUCT |
2009 | |
1388 | 2010 class PrintReachableOopClosure: public OopClosure { |
342 | 2011 private: |
2012 G1CollectedHeap* _g1h; | |
2013 CMBitMapRO* _bitmap; | |
2014 outputStream* _out; | |
1044 | 2015 bool _use_prev_marking; |
1388 | 2016 bool _all; |
342 | 2017 |
2018 public: | |
1388 | 2019 PrintReachableOopClosure(CMBitMapRO* bitmap, |
2020 outputStream* out, | |
2021 bool use_prev_marking, | |
2022 bool all) : | |
1044 | 2023 _g1h(G1CollectedHeap::heap()), |
1388 | 2024 _bitmap(bitmap), _out(out), _use_prev_marking(use_prev_marking), _all(all) { } |
342 | 2025 |
845
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
2026 void do_oop(narrowOop* p) { do_oop_work(p); } |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
2027 void do_oop( oop* p) { do_oop_work(p); } |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
2028 |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
2029 template <class T> void do_oop_work(T* p) { |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
2030 oop obj = oopDesc::load_decode_heap_oop(p); |
342 | 2031 const char* str = NULL; |
2032 const char* str2 = ""; | |
2033 | |
1388 | 2034 if (obj == NULL) { |
2035 str = ""; | |
2036 } else if (!_g1h->is_in_g1_reserved(obj)) { | |
2037 str = " O"; | |
2038 } else { | |
342 | 2039 HeapRegion* hr = _g1h->heap_region_containing(obj); |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
2040 guarantee(hr != NULL, "invariant"); |
1044 | 2041 bool over_tams = false; |
2042 if (_use_prev_marking) { | |
2043 over_tams = hr->obj_allocated_since_prev_marking(obj); | |
2044 } else { | |
2045 over_tams = hr->obj_allocated_since_next_marking(obj); | |
2046 } | |
1388 | 2047 bool marked = _bitmap->isMarked((HeapWord*) obj); |
1044 | 2048 |
2049 if (over_tams) { | |
1388 | 2050 str = " >"; |
2051 if (marked) { | |
342 | 2052 str2 = " AND MARKED"; |
1044 | 2053 } |
1388 | 2054 } else if (marked) { |
2055 str = " M"; | |
1044 | 2056 } else { |
1388 | 2057 str = " NOT"; |
1044 | 2058 } |
342 | 2059 } |
2060 | |
1388 | 2061 _out->print_cr(" "PTR_FORMAT": "PTR_FORMAT"%s%s", |
342 | 2062 p, (void*) obj, str, str2); |
2063 } | |
2064 }; | |
2065 | |
1388 | 2066 class PrintReachableObjectClosure : public ObjectClosure { |
342 | 2067 private: |
1044 | 2068 CMBitMapRO* _bitmap; |
342 | 2069 outputStream* _out; |
1044 | 2070 bool _use_prev_marking; |
1388 | 2071 bool _all; |
2072 HeapRegion* _hr; | |
342 | 2073 |
2074 public: | |
1388 | 2075 PrintReachableObjectClosure(CMBitMapRO* bitmap, |
2076 outputStream* out, | |
2077 bool use_prev_marking, | |
2078 bool all, | |
2079 HeapRegion* hr) : | |
2080 _bitmap(bitmap), _out(out), | |
2081 _use_prev_marking(use_prev_marking), _all(all), _hr(hr) { } | |
2082 | |
2083 void do_object(oop o) { | |
2084 bool over_tams; | |
2085 if (_use_prev_marking) { | |
2086 over_tams = _hr->obj_allocated_since_prev_marking(o); | |
2087 } else { | |
2088 over_tams = _hr->obj_allocated_since_next_marking(o); | |
2089 } | |
2090 bool marked = _bitmap->isMarked((HeapWord*) o); | |
2091 bool print_it = _all || over_tams || marked; | |
2092 | |
2093 if (print_it) { | |
2094 _out->print_cr(" "PTR_FORMAT"%s", | |
2095 o, (over_tams) ? " >" : (marked) ? " M" : ""); | |
2096 PrintReachableOopClosure oopCl(_bitmap, _out, _use_prev_marking, _all); | |
2097 o->oop_iterate(&oopCl); | |
2098 } | |
342 | 2099 } |
2100 }; | |
2101 | |
1388 | 2102 class PrintReachableRegionClosure : public HeapRegionClosure { |
342 | 2103 private: |
1044 | 2104 CMBitMapRO* _bitmap; |
342 | 2105 outputStream* _out; |
1044 | 2106 bool _use_prev_marking; |
1388 | 2107 bool _all; |
342 | 2108 |
2109 public: | |
2110 bool doHeapRegion(HeapRegion* hr) { | |
2111 HeapWord* b = hr->bottom(); | |
2112 HeapWord* e = hr->end(); | |
2113 HeapWord* t = hr->top(); | |
1044 | 2114 HeapWord* p = NULL; |
2115 if (_use_prev_marking) { | |
2116 p = hr->prev_top_at_mark_start(); | |
2117 } else { | |
2118 p = hr->next_top_at_mark_start(); | |
2119 } | |
342 | 2120 _out->print_cr("** ["PTR_FORMAT", "PTR_FORMAT"] top: "PTR_FORMAT" " |
1044 | 2121 "TAMS: "PTR_FORMAT, b, e, t, p); |
1388 | 2122 _out->cr(); |
2123 | |
2124 HeapWord* from = b; | |
2125 HeapWord* to = t; | |
2126 | |
2127 if (to > from) { | |
2128 _out->print_cr("Objects in ["PTR_FORMAT", "PTR_FORMAT"]", from, to); | |
2129 _out->cr(); | |
2130 PrintReachableObjectClosure ocl(_bitmap, _out, | |
2131 _use_prev_marking, _all, hr); | |
2132 hr->object_iterate_mem_careful(MemRegion(from, to), &ocl); | |
2133 _out->cr(); | |
2134 } | |
342 | 2135 |
2136 return false; | |
2137 } | |
2138 | |
1388 | 2139 PrintReachableRegionClosure(CMBitMapRO* bitmap, |
2140 outputStream* out, | |
2141 bool use_prev_marking, | |
2142 bool all) : | |
2143 _bitmap(bitmap), _out(out), _use_prev_marking(use_prev_marking), _all(all) { } | |
342 | 2144 }; |
2145 | |
1388 | 2146 void ConcurrentMark::print_reachable(const char* str, |
2147 bool use_prev_marking, | |
2148 bool all) { | |
2149 gclog_or_tty->cr(); | |
2150 gclog_or_tty->print_cr("== Doing heap dump... "); | |
1044 | 2151 |
2152 if (G1PrintReachableBaseFile == NULL) { | |
2153 gclog_or_tty->print_cr(" #### error: no base file defined"); | |
2154 return; | |
2155 } | |
2156 | |
2157 if (strlen(G1PrintReachableBaseFile) + 1 + strlen(str) > | |
2158 (JVM_MAXPATHLEN - 1)) { | |
2159 gclog_or_tty->print_cr(" #### error: file name too long"); | |
2160 return; | |
2161 } | |
2162 | |
2163 char file_name[JVM_MAXPATHLEN]; | |
2164 sprintf(file_name, "%s.%s", G1PrintReachableBaseFile, str); | |
2165 gclog_or_tty->print_cr(" dumping to file %s", file_name); | |
2166 | |
2167 fileStream fout(file_name); | |
2168 if (!fout.is_open()) { | |
2169 gclog_or_tty->print_cr(" #### error: could not open file"); | |
2170 return; | |
2171 } | |
2172 | |
2173 outputStream* out = &fout; | |
2174 | |
2175 CMBitMapRO* bitmap = NULL; | |
2176 if (use_prev_marking) { | |
2177 bitmap = _prevMarkBitMap; | |
2178 } else { | |
2179 bitmap = _nextMarkBitMap; | |
2180 } | |
2181 | |
2182 out->print_cr("-- USING %s", (use_prev_marking) ? "PTAMS" : "NTAMS"); | |
2183 out->cr(); | |
2184 | |
1388 | 2185 out->print_cr("--- ITERATING OVER REGIONS"); |
1044 | 2186 out->cr(); |
1388 | 2187 PrintReachableRegionClosure rcl(bitmap, out, use_prev_marking, all); |
2188 _g1h->heap_region_iterate(&rcl); | |
1044 | 2189 out->cr(); |
2190 | |
2191 gclog_or_tty->print_cr(" done"); | |
1388 | 2192 gclog_or_tty->flush(); |
342 | 2193 } |
2194 | |
1044 | 2195 #endif // PRODUCT |
2196 | |
342 | 2197 // This note is for drainAllSATBBuffers and the code in between. |
2198 // In the future we could reuse a task to do this work during an | |
2199 // evacuation pause (since now tasks are not active and can be claimed | |
2200 // during an evacuation pause). This was a late change to the code and | |
2201 // is currently not being taken advantage of. | |
2202 | |
2203 class CMGlobalObjectClosure : public ObjectClosure { | |
2204 private: | |
2205 ConcurrentMark* _cm; | |
2206 | |
2207 public: | |
2208 void do_object(oop obj) { | |
2209 _cm->deal_with_reference(obj); | |
2210 } | |
2211 | |
2212 CMGlobalObjectClosure(ConcurrentMark* cm) : _cm(cm) { } | |
2213 }; | |
2214 | |
2215 void ConcurrentMark::deal_with_reference(oop obj) { | |
2216 if (verbose_high()) | |
2217 gclog_or_tty->print_cr("[global] we're dealing with reference "PTR_FORMAT, | |
2218 (void*) obj); | |
2219 | |
2220 | |
2221 HeapWord* objAddr = (HeapWord*) obj; | |
845
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
2222 assert(obj->is_oop_or_null(true /* ignore mark word */), "Error"); |
342 | 2223 if (_g1h->is_in_g1_reserved(objAddr)) { |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
2224 assert(obj != NULL, "is_in_g1_reserved should ensure this"); |
342 | 2225 HeapRegion* hr = _g1h->heap_region_containing(obj); |
2226 if (_g1h->is_obj_ill(obj, hr)) { | |
2227 if (verbose_high()) | |
2228 gclog_or_tty->print_cr("[global] "PTR_FORMAT" is not considered " | |
2229 "marked", (void*) obj); | |
2230 | |
2231 // we need to mark it first | |
2232 if (_nextMarkBitMap->parMark(objAddr)) { | |
2233 // No OrderAccess:store_load() is needed. It is implicit in the | |
2234 // CAS done in parMark(objAddr) above | |
2235 HeapWord* finger = _finger; | |
2236 if (objAddr < finger) { | |
2237 if (verbose_high()) | |
2238 gclog_or_tty->print_cr("[global] below the global finger " | |
2239 "("PTR_FORMAT"), pushing it", finger); | |
2240 if (!mark_stack_push(obj)) { | |
2241 if (verbose_low()) | |
2242 gclog_or_tty->print_cr("[global] global stack overflow during " | |
2243 "deal_with_reference"); | |
2244 } | |
2245 } | |
2246 } | |
2247 } | |
2248 } | |
2249 } | |
2250 | |
2251 void ConcurrentMark::drainAllSATBBuffers() { | |
2252 CMGlobalObjectClosure oc(this); | |
2253 SATBMarkQueueSet& satb_mq_set = JavaThread::satb_mark_queue_set(); | |
2254 satb_mq_set.set_closure(&oc); | |
2255 | |
2256 while (satb_mq_set.apply_closure_to_completed_buffer()) { | |
2257 if (verbose_medium()) | |
2258 gclog_or_tty->print_cr("[global] processed an SATB buffer"); | |
2259 } | |
2260 | |
2261 // no need to check whether we should do this, as this is only | |
2262 // called during an evacuation pause | |
2263 satb_mq_set.iterate_closure_all_threads(); | |
2264 | |
2265 satb_mq_set.set_closure(NULL); | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
2266 assert(satb_mq_set.completed_buffers_num() == 0, "invariant"); |
342 | 2267 } |
2268 | |
2269 void ConcurrentMark::markPrev(oop p) { | |
2270 // Note we are overriding the read-only view of the prev map here, via | |
2271 // the cast. | |
2272 ((CMBitMap*)_prevMarkBitMap)->mark((HeapWord*)p); | |
2273 } | |
2274 | |
2275 void ConcurrentMark::clear(oop p) { | |
2276 assert(p != NULL && p->is_oop(), "expected an oop"); | |
2277 HeapWord* addr = (HeapWord*)p; | |
2278 assert(addr >= _nextMarkBitMap->startWord() || | |
2279 addr < _nextMarkBitMap->endWord(), "in a region"); | |
2280 | |
2281 _nextMarkBitMap->clear(addr); | |
2282 } | |
2283 | |
2284 void ConcurrentMark::clearRangeBothMaps(MemRegion mr) { | |
2285 // Note we are overriding the read-only view of the prev map here, via | |
2286 // the cast. | |
2287 ((CMBitMap*)_prevMarkBitMap)->clearRange(mr); | |
2288 _nextMarkBitMap->clearRange(mr); | |
2289 } | |
2290 | |
2291 HeapRegion* | |
2292 ConcurrentMark::claim_region(int task_num) { | |
2293 // "checkpoint" the finger | |
2294 HeapWord* finger = _finger; | |
2295 | |
2296 // _heap_end will not change underneath our feet; it only changes at | |
2297 // yield points. | |
2298 while (finger < _heap_end) { | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
2299 assert(_g1h->is_in_g1_reserved(finger), "invariant"); |
342 | 2300 |
2301 // is the gap between reading the finger and doing the CAS too long? | |
2302 | |
2303 HeapRegion* curr_region = _g1h->heap_region_containing(finger); | |
2304 HeapWord* bottom = curr_region->bottom(); | |
2305 HeapWord* end = curr_region->end(); | |
2306 HeapWord* limit = curr_region->next_top_at_mark_start(); | |
2307 | |
2308 if (verbose_low()) | |
2309 gclog_or_tty->print_cr("[%d] curr_region = "PTR_FORMAT" " | |
2310 "["PTR_FORMAT", "PTR_FORMAT"), " | |
2311 "limit = "PTR_FORMAT, | |
2312 task_num, curr_region, bottom, end, limit); | |
2313 | |
2314 HeapWord* res = | |
2315 (HeapWord*) Atomic::cmpxchg_ptr(end, &_finger, finger); | |
2316 if (res == finger) { | |
2317 // we succeeded | |
2318 | |
2319 // notice that _finger == end cannot be guaranteed here since, | |
2320 // someone else might have moved the finger even further | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
2321 assert(_finger >= end, "the finger should have moved forward"); |
342 | 2322 |
2323 if (verbose_low()) | |
2324 gclog_or_tty->print_cr("[%d] we were successful with region = " | |
2325 PTR_FORMAT, task_num, curr_region); | |
2326 | |
2327 if (limit > bottom) { | |
2328 if (verbose_low()) | |
2329 gclog_or_tty->print_cr("[%d] region "PTR_FORMAT" is not empty, " | |
2330 "returning it ", task_num, curr_region); | |
2331 return curr_region; | |
2332 } else { | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
2333 assert(limit == bottom, |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
2334 "the region limit should be at bottom"); |
342 | 2335 if (verbose_low()) |
2336 gclog_or_tty->print_cr("[%d] region "PTR_FORMAT" is empty, " | |
2337 "returning NULL", task_num, curr_region); | |
2338 // we return NULL and the caller should try calling | |
2339 // claim_region() again. | |
2340 return NULL; | |
2341 } | |
2342 } else { | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
2343 assert(_finger > finger, "the finger should have moved forward"); |
342 | 2344 if (verbose_low()) |
2345 gclog_or_tty->print_cr("[%d] somebody else moved the finger, " | |
2346 "global finger = "PTR_FORMAT", " | |
2347 "our finger = "PTR_FORMAT, | |
2348 task_num, _finger, finger); | |
2349 | |
2350 // read it again | |
2351 finger = _finger; | |
2352 } | |
2353 } | |
2354 | |
2355 return NULL; | |
2356 } | |
2357 | |
1835
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
2358 bool ConcurrentMark::invalidate_aborted_regions_in_cset() { |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
2359 bool result = false; |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
2360 for (int i = 0; i < (int)_max_task_num; ++i) { |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
2361 CMTask* the_task = _tasks[i]; |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
2362 MemRegion mr = the_task->aborted_region(); |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
2363 if (mr.start() != NULL) { |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
2364 assert(mr.end() != NULL, "invariant"); |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
2365 assert(mr.word_size() > 0, "invariant"); |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
2366 HeapRegion* hr = _g1h->heap_region_containing(mr.start()); |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
2367 assert(hr != NULL, "invariant"); |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
2368 if (hr->in_collection_set()) { |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
2369 // The region points into the collection set |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
2370 the_task->set_aborted_region(MemRegion()); |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
2371 result = true; |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
2372 } |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
2373 } |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
2374 } |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
2375 return result; |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
2376 } |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
2377 |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
2378 bool ConcurrentMark::has_aborted_regions() { |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
2379 for (int i = 0; i < (int)_max_task_num; ++i) { |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
2380 CMTask* the_task = _tasks[i]; |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
2381 MemRegion mr = the_task->aborted_region(); |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
2382 if (mr.start() != NULL) { |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
2383 assert(mr.end() != NULL, "invariant"); |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
2384 assert(mr.word_size() > 0, "invariant"); |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
2385 return true; |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
2386 } |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
2387 } |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
2388 return false; |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
2389 } |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
2390 |
342 | 2391 void ConcurrentMark::oops_do(OopClosure* cl) { |
2392 if (_markStack.size() > 0 && verbose_low()) | |
2393 gclog_or_tty->print_cr("[global] scanning the global marking stack, " | |
2394 "size = %d", _markStack.size()); | |
2395 // we first iterate over the contents of the mark stack... | |
2396 _markStack.oops_do(cl); | |
2397 | |
2398 for (int i = 0; i < (int)_max_task_num; ++i) { | |
2399 OopTaskQueue* queue = _task_queues->queue((int)i); | |
2400 | |
2401 if (queue->size() > 0 && verbose_low()) | |
2402 gclog_or_tty->print_cr("[global] scanning task queue of task %d, " | |
2403 "size = %d", i, queue->size()); | |
2404 | |
2405 // ...then over the contents of the all the task queues. | |
2406 queue->oops_do(cl); | |
2407 } | |
2408 | |
1835
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
2409 // Invalidate any entries, that are in the region stack, that |
342 | 2410 // point into the collection set |
2411 if (_regionStack.invalidate_entries_into_cset()) { | |
2412 // otherwise, any gray objects copied during the evacuation pause | |
2413 // might not be visited. | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
2414 assert(_should_gray_objects, "invariant"); |
342 | 2415 } |
1835
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
2416 |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
2417 // Invalidate any aborted regions, recorded in the individual CM |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
2418 // tasks, that point into the collection set. |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
2419 if (invalidate_aborted_regions_in_cset()) { |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
2420 // otherwise, any gray objects copied during the evacuation pause |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
2421 // might not be visited. |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
2422 assert(_should_gray_objects, "invariant"); |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
2423 } |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
2424 |
342 | 2425 } |
2426 | |
2427 void ConcurrentMark::clear_marking_state() { | |
2428 _markStack.setEmpty(); | |
2429 _markStack.clear_overflow(); | |
2430 _regionStack.setEmpty(); | |
2431 _regionStack.clear_overflow(); | |
2432 clear_has_overflown(); | |
2433 _finger = _heap_start; | |
2434 | |
2435 for (int i = 0; i < (int)_max_task_num; ++i) { | |
2436 OopTaskQueue* queue = _task_queues->queue(i); | |
2437 queue->set_empty(); | |
1885
a5c514e74487
6988458: G1: assert(mr.end() <= _cm->finger()) failed: otherwise the region shouldn't be on the stack
johnc
parents:
1835
diff
changeset
|
2438 // Clear any partial regions from the CMTasks |
a5c514e74487
6988458: G1: assert(mr.end() <= _cm->finger()) failed: otherwise the region shouldn't be on the stack
johnc
parents:
1835
diff
changeset
|
2439 _tasks[i]->clear_aborted_region(); |
342 | 2440 } |
2441 } | |
2442 | |
2443 void ConcurrentMark::print_stats() { | |
2444 if (verbose_stats()) { | |
2445 gclog_or_tty->print_cr("---------------------------------------------------------------------"); | |
2446 for (size_t i = 0; i < _active_tasks; ++i) { | |
2447 _tasks[i]->print_stats(); | |
2448 gclog_or_tty->print_cr("---------------------------------------------------------------------"); | |
2449 } | |
2450 } | |
2451 } | |
2452 | |
2453 class CSMarkOopClosure: public OopClosure { | |
2454 friend class CSMarkBitMapClosure; | |
2455 | |
2456 G1CollectedHeap* _g1h; | |
2457 CMBitMap* _bm; | |
2458 ConcurrentMark* _cm; | |
2459 oop* _ms; | |
2460 jint* _array_ind_stack; | |
2461 int _ms_size; | |
2462 int _ms_ind; | |
2463 int _array_increment; | |
2464 | |
2465 bool push(oop obj, int arr_ind = 0) { | |
2466 if (_ms_ind == _ms_size) { | |
2467 gclog_or_tty->print_cr("Mark stack is full."); | |
2468 return false; | |
2469 } | |
2470 _ms[_ms_ind] = obj; | |
2471 if (obj->is_objArray()) _array_ind_stack[_ms_ind] = arr_ind; | |
2472 _ms_ind++; | |
2473 return true; | |
2474 } | |
2475 | |
2476 oop pop() { | |
2477 if (_ms_ind == 0) return NULL; | |
2478 else { | |
2479 _ms_ind--; | |
2480 return _ms[_ms_ind]; | |
2481 } | |
2482 } | |
2483 | |
845
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
2484 template <class T> bool drain() { |
342 | 2485 while (_ms_ind > 0) { |
2486 oop obj = pop(); | |
2487 assert(obj != NULL, "Since index was non-zero."); | |
2488 if (obj->is_objArray()) { | |
2489 jint arr_ind = _array_ind_stack[_ms_ind]; | |
2490 objArrayOop aobj = objArrayOop(obj); | |
2491 jint len = aobj->length(); | |
2492 jint next_arr_ind = arr_ind + _array_increment; | |
2493 if (next_arr_ind < len) { | |
2494 push(obj, next_arr_ind); | |
2495 } | |
2496 // Now process this portion of this one. | |
2497 int lim = MIN2(next_arr_ind, len); | |
2498 for (int j = arr_ind; j < lim; j++) { | |
912
308762b2bf14
6872000: G1: compilation fails on linux/older gcc
apetrusenko
parents:
866
diff
changeset
|
2499 do_oop(aobj->objArrayOopDesc::obj_at_addr<T>(j)); |
342 | 2500 } |
2501 | |
2502 } else { | |
2503 obj->oop_iterate(this); | |
2504 } | |
2505 if (abort()) return false; | |
2506 } | |
2507 return true; | |
2508 } | |
2509 | |
2510 public: | |
2511 CSMarkOopClosure(ConcurrentMark* cm, int ms_size) : | |
2512 _g1h(G1CollectedHeap::heap()), | |
2513 _cm(cm), | |
2514 _bm(cm->nextMarkBitMap()), | |
2515 _ms_size(ms_size), _ms_ind(0), | |
2516 _ms(NEW_C_HEAP_ARRAY(oop, ms_size)), | |
2517 _array_ind_stack(NEW_C_HEAP_ARRAY(jint, ms_size)), | |
2518 _array_increment(MAX2(ms_size/8, 16)) | |
2519 {} | |
2520 | |
2521 ~CSMarkOopClosure() { | |
2522 FREE_C_HEAP_ARRAY(oop, _ms); | |
2523 FREE_C_HEAP_ARRAY(jint, _array_ind_stack); | |
2524 } | |
2525 | |
845
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
2526 virtual void do_oop(narrowOop* p) { do_oop_work(p); } |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
2527 virtual void do_oop( oop* p) { do_oop_work(p); } |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
2528 |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
2529 template <class T> void do_oop_work(T* p) { |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
2530 T heap_oop = oopDesc::load_heap_oop(p); |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
2531 if (oopDesc::is_null(heap_oop)) return; |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
2532 oop obj = oopDesc::decode_heap_oop_not_null(heap_oop); |
342 | 2533 if (obj->is_forwarded()) { |
2534 // If the object has already been forwarded, we have to make sure | |
2535 // that it's marked. So follow the forwarding pointer. Note that | |
2536 // this does the right thing for self-forwarding pointers in the | |
2537 // evacuation failure case. | |
2538 obj = obj->forwardee(); | |
2539 } | |
2540 HeapRegion* hr = _g1h->heap_region_containing(obj); | |
2541 if (hr != NULL) { | |
2542 if (hr->in_collection_set()) { | |
2543 if (_g1h->is_obj_ill(obj)) { | |
2544 _bm->mark((HeapWord*)obj); | |
2545 if (!push(obj)) { | |
2546 gclog_or_tty->print_cr("Setting abort in CSMarkOopClosure because push failed."); | |
2547 set_abort(); | |
2548 } | |
2549 } | |
2550 } else { | |
2551 // Outside the collection set; we need to gray it | |
2552 _cm->deal_with_reference(obj); | |
2553 } | |
2554 } | |
2555 } | |
2556 }; | |
2557 | |
2558 class CSMarkBitMapClosure: public BitMapClosure { | |
2559 G1CollectedHeap* _g1h; | |
2560 CMBitMap* _bitMap; | |
2561 ConcurrentMark* _cm; | |
2562 CSMarkOopClosure _oop_cl; | |
2563 public: | |
2564 CSMarkBitMapClosure(ConcurrentMark* cm, int ms_size) : | |
2565 _g1h(G1CollectedHeap::heap()), | |
2566 _bitMap(cm->nextMarkBitMap()), | |
2567 _oop_cl(cm, ms_size) | |
2568 {} | |
2569 | |
2570 ~CSMarkBitMapClosure() {} | |
2571 | |
2572 bool do_bit(size_t offset) { | |
2573 // convert offset into a HeapWord* | |
2574 HeapWord* addr = _bitMap->offsetToHeapWord(offset); | |
2575 assert(_bitMap->endWord() && addr < _bitMap->endWord(), | |
2576 "address out of range"); | |
2577 assert(_bitMap->isMarked(addr), "tautology"); | |
2578 oop obj = oop(addr); | |
2579 if (!obj->is_forwarded()) { | |
2580 if (!_oop_cl.push(obj)) return false; | |
845
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
2581 if (UseCompressedOops) { |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
2582 if (!_oop_cl.drain<narrowOop>()) return false; |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
2583 } else { |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
2584 if (!_oop_cl.drain<oop>()) return false; |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
2585 } |
342 | 2586 } |
2587 // Otherwise... | |
2588 return true; | |
2589 } | |
2590 }; | |
2591 | |
2592 | |
2593 class CompleteMarkingInCSHRClosure: public HeapRegionClosure { | |
2594 CMBitMap* _bm; | |
2595 CSMarkBitMapClosure _bit_cl; | |
2596 enum SomePrivateConstants { | |
2597 MSSize = 1000 | |
2598 }; | |
2599 bool _completed; | |
2600 public: | |
2601 CompleteMarkingInCSHRClosure(ConcurrentMark* cm) : | |
2602 _bm(cm->nextMarkBitMap()), | |
2603 _bit_cl(cm, MSSize), | |
2604 _completed(true) | |
2605 {} | |
2606 | |
2607 ~CompleteMarkingInCSHRClosure() {} | |
2608 | |
2609 bool doHeapRegion(HeapRegion* r) { | |
2610 if (!r->evacuation_failed()) { | |
2611 MemRegion mr = MemRegion(r->bottom(), r->next_top_at_mark_start()); | |
2612 if (!mr.is_empty()) { | |
2613 if (!_bm->iterate(&_bit_cl, mr)) { | |
2614 _completed = false; | |
2615 return true; | |
2616 } | |
2617 } | |
2618 } | |
2619 return false; | |
2620 } | |
2621 | |
2622 bool completed() { return _completed; } | |
2623 }; | |
2624 | |
2625 class ClearMarksInHRClosure: public HeapRegionClosure { | |
2626 CMBitMap* _bm; | |
2627 public: | |
2628 ClearMarksInHRClosure(CMBitMap* bm): _bm(bm) { } | |
2629 | |
2630 bool doHeapRegion(HeapRegion* r) { | |
2631 if (!r->used_region().is_empty() && !r->evacuation_failed()) { | |
2632 MemRegion usedMR = r->used_region(); | |
2633 _bm->clearRange(r->used_region()); | |
2634 } | |
2635 return false; | |
2636 } | |
2637 }; | |
2638 | |
2639 void ConcurrentMark::complete_marking_in_collection_set() { | |
2640 G1CollectedHeap* g1h = G1CollectedHeap::heap(); | |
2641 | |
2642 if (!g1h->mark_in_progress()) { | |
2643 g1h->g1_policy()->record_mark_closure_time(0.0); | |
2644 return; | |
2645 } | |
2646 | |
2647 int i = 1; | |
2648 double start = os::elapsedTime(); | |
2649 while (true) { | |
2650 i++; | |
2651 CompleteMarkingInCSHRClosure cmplt(this); | |
2652 g1h->collection_set_iterate(&cmplt); | |
2653 if (cmplt.completed()) break; | |
2654 } | |
2655 double end_time = os::elapsedTime(); | |
2656 double elapsed_time_ms = (end_time - start) * 1000.0; | |
2657 g1h->g1_policy()->record_mark_closure_time(elapsed_time_ms); | |
2658 | |
2659 ClearMarksInHRClosure clr(nextMarkBitMap()); | |
2660 g1h->collection_set_iterate(&clr); | |
2661 } | |
2662 | |
2663 // The next two methods deal with the following optimisation. Some | |
2664 // objects are gray by being marked and located above the finger. If | |
2665 // they are copied, during an evacuation pause, below the finger then | |
2666 // the need to be pushed on the stack. The observation is that, if | |
2667 // there are no regions in the collection set located above the | |
2668 // finger, then the above cannot happen, hence we do not need to | |
2669 // explicitly gray any objects when copying them to below the | |
2670 // finger. The global stack will be scanned to ensure that, if it | |
2671 // points to objects being copied, it will update their | |
2672 // location. There is a tricky situation with the gray objects in | |
2673 // region stack that are being coped, however. See the comment in | |
2674 // newCSet(). | |
2675 | |
2676 void ConcurrentMark::newCSet() { | |
2677 if (!concurrent_marking_in_progress()) | |
2678 // nothing to do if marking is not in progress | |
2679 return; | |
2680 | |
2681 // find what the lowest finger is among the global and local fingers | |
2682 _min_finger = _finger; | |
2683 for (int i = 0; i < (int)_max_task_num; ++i) { | |
2684 CMTask* task = _tasks[i]; | |
2685 HeapWord* task_finger = task->finger(); | |
2686 if (task_finger != NULL && task_finger < _min_finger) | |
2687 _min_finger = task_finger; | |
2688 } | |
2689 | |
2690 _should_gray_objects = false; | |
2691 | |
2692 // This fixes a very subtle and fustrating bug. It might be the case | |
2693 // that, during en evacuation pause, heap regions that contain | |
2694 // objects that are gray (by being in regions contained in the | |
2695 // region stack) are included in the collection set. Since such gray | |
2696 // objects will be moved, and because it's not easy to redirect | |
2697 // region stack entries to point to a new location (because objects | |
2698 // in one region might be scattered to multiple regions after they | |
2699 // are copied), one option is to ensure that all marked objects | |
2700 // copied during a pause are pushed on the stack. Notice, however, | |
2701 // that this problem can only happen when the region stack is not | |
2702 // empty during an evacuation pause. So, we make the fix a bit less | |
2703 // conservative and ensure that regions are pushed on the stack, | |
2704 // irrespective whether all collection set regions are below the | |
2705 // finger, if the region stack is not empty. This is expected to be | |
2706 // a rare case, so I don't think it's necessary to be smarted about it. | |
1835
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
2707 if (!region_stack_empty() || has_aborted_regions()) |
342 | 2708 _should_gray_objects = true; |
2709 } | |
2710 | |
2711 void ConcurrentMark::registerCSetRegion(HeapRegion* hr) { | |
2712 if (!concurrent_marking_in_progress()) | |
2713 return; | |
2714 | |
2715 HeapWord* region_end = hr->end(); | |
2716 if (region_end > _min_finger) | |
2717 _should_gray_objects = true; | |
2718 } | |
2719 | |
2720 // abandon current marking iteration due to a Full GC | |
2721 void ConcurrentMark::abort() { | |
2722 // Clear all marks to force marking thread to do nothing | |
2723 _nextMarkBitMap->clearAll(); | |
2724 // Empty mark stack | |
2725 clear_marking_state(); | |
1835
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
2726 for (int i = 0; i < (int)_max_task_num; ++i) { |
342 | 2727 _tasks[i]->clear_region_fields(); |
1835
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
2728 } |
342 | 2729 _has_aborted = true; |
2730 | |
2731 SATBMarkQueueSet& satb_mq_set = JavaThread::satb_mark_queue_set(); | |
2732 satb_mq_set.abandon_partial_marking(); | |
1317
d4197f8d516a
6935821: G1: threads created during marking do not active their SATB queues
tonyp
parents:
1314
diff
changeset
|
2733 // This can be called either during or outside marking, we'll read |
d4197f8d516a
6935821: G1: threads created during marking do not active their SATB queues
tonyp
parents:
1314
diff
changeset
|
2734 // the expected_active value from the SATB queue set. |
d4197f8d516a
6935821: G1: threads created during marking do not active their SATB queues
tonyp
parents:
1314
diff
changeset
|
2735 satb_mq_set.set_active_all_threads( |
d4197f8d516a
6935821: G1: threads created during marking do not active their SATB queues
tonyp
parents:
1314
diff
changeset
|
2736 false, /* new active value */ |
d4197f8d516a
6935821: G1: threads created during marking do not active their SATB queues
tonyp
parents:
1314
diff
changeset
|
2737 satb_mq_set.is_active() /* expected_active */); |
342 | 2738 } |
2739 | |
2740 static void print_ms_time_info(const char* prefix, const char* name, | |
2741 NumberSeq& ns) { | |
2742 gclog_or_tty->print_cr("%s%5d %12s: total time = %8.2f s (avg = %8.2f ms).", | |
2743 prefix, ns.num(), name, ns.sum()/1000.0, ns.avg()); | |
2744 if (ns.num() > 0) { | |
2745 gclog_or_tty->print_cr("%s [std. dev = %8.2f ms, max = %8.2f ms]", | |
2746 prefix, ns.sd(), ns.maximum()); | |
2747 } | |
2748 } | |
2749 | |
2750 void ConcurrentMark::print_summary_info() { | |
2751 gclog_or_tty->print_cr(" Concurrent marking:"); | |
2752 print_ms_time_info(" ", "init marks", _init_times); | |
2753 print_ms_time_info(" ", "remarks", _remark_times); | |
2754 { | |
2755 print_ms_time_info(" ", "final marks", _remark_mark_times); | |
2756 print_ms_time_info(" ", "weak refs", _remark_weak_ref_times); | |
2757 | |
2758 } | |
2759 print_ms_time_info(" ", "cleanups", _cleanup_times); | |
2760 gclog_or_tty->print_cr(" Final counting total time = %8.2f s (avg = %8.2f ms).", | |
2761 _total_counting_time, | |
2762 (_cleanup_times.num() > 0 ? _total_counting_time * 1000.0 / | |
2763 (double)_cleanup_times.num() | |
2764 : 0.0)); | |
2765 if (G1ScrubRemSets) { | |
2766 gclog_or_tty->print_cr(" RS scrub total time = %8.2f s (avg = %8.2f ms).", | |
2767 _total_rs_scrub_time, | |
2768 (_cleanup_times.num() > 0 ? _total_rs_scrub_time * 1000.0 / | |
2769 (double)_cleanup_times.num() | |
2770 : 0.0)); | |
2771 } | |
2772 gclog_or_tty->print_cr(" Total stop_world time = %8.2f s.", | |
2773 (_init_times.sum() + _remark_times.sum() + | |
2774 _cleanup_times.sum())/1000.0); | |
2775 gclog_or_tty->print_cr(" Total concurrent time = %8.2f s " | |
2776 "(%8.2f s marking, %8.2f s counting).", | |
2777 cmThread()->vtime_accum(), | |
2778 cmThread()->vtime_mark_accum(), | |
2779 cmThread()->vtime_count_accum()); | |
2780 } | |
2781 | |
1019 | 2782 void ConcurrentMark::print_worker_threads_on(outputStream* st) const { |
2783 _parallel_workers->print_worker_threads_on(st); | |
2784 } | |
2785 | |
342 | 2786 // Closures |
2787 // XXX: there seems to be a lot of code duplication here; | |
2788 // should refactor and consolidate the shared code. | |
2789 | |
2790 // This closure is used to mark refs into the CMS generation in | |
2791 // the CMS bit map. Called at the first checkpoint. | |
2792 | |
2793 // We take a break if someone is trying to stop the world. | |
2794 bool ConcurrentMark::do_yield_check(int worker_i) { | |
2795 if (should_yield()) { | |
2796 if (worker_i == 0) | |
2797 _g1h->g1_policy()->record_concurrent_pause(); | |
2798 cmThread()->yield(); | |
2799 if (worker_i == 0) | |
2800 _g1h->g1_policy()->record_concurrent_pause_end(); | |
2801 return true; | |
2802 } else { | |
2803 return false; | |
2804 } | |
2805 } | |
2806 | |
2807 bool ConcurrentMark::should_yield() { | |
2808 return cmThread()->should_yield(); | |
2809 } | |
2810 | |
2811 bool ConcurrentMark::containing_card_is_marked(void* p) { | |
2812 size_t offset = pointer_delta(p, _g1h->reserved_region().start(), 1); | |
2813 return _card_bm.at(offset >> CardTableModRefBS::card_shift); | |
2814 } | |
2815 | |
2816 bool ConcurrentMark::containing_cards_are_marked(void* start, | |
2817 void* last) { | |
2818 return | |
2819 containing_card_is_marked(start) && | |
2820 containing_card_is_marked(last); | |
2821 } | |
2822 | |
2823 #ifndef PRODUCT | |
2824 // for debugging purposes | |
2825 void ConcurrentMark::print_finger() { | |
2826 gclog_or_tty->print_cr("heap ["PTR_FORMAT", "PTR_FORMAT"), global finger = "PTR_FORMAT, | |
2827 _heap_start, _heap_end, _finger); | |
2828 for (int i = 0; i < (int) _max_task_num; ++i) { | |
2829 gclog_or_tty->print(" %d: "PTR_FORMAT, i, _tasks[i]->finger()); | |
2830 } | |
2831 gclog_or_tty->print_cr(""); | |
2832 } | |
2833 #endif | |
2834 | |
2835 // Closure for iteration over bitmaps | |
2836 class CMBitMapClosure : public BitMapClosure { | |
2837 private: | |
2838 // the bitmap that is being iterated over | |
2839 CMBitMap* _nextMarkBitMap; | |
2840 ConcurrentMark* _cm; | |
2841 CMTask* _task; | |
2842 // true if we're scanning a heap region claimed by the task (so that | |
2843 // we move the finger along), false if we're not, i.e. currently when | |
2844 // scanning a heap region popped from the region stack (so that we | |
2845 // do not move the task finger along; it'd be a mistake if we did so). | |
2846 bool _scanning_heap_region; | |
2847 | |
2848 public: | |
2849 CMBitMapClosure(CMTask *task, | |
2850 ConcurrentMark* cm, | |
2851 CMBitMap* nextMarkBitMap) | |
2852 : _task(task), _cm(cm), _nextMarkBitMap(nextMarkBitMap) { } | |
2853 | |
2854 void set_scanning_heap_region(bool scanning_heap_region) { | |
2855 _scanning_heap_region = scanning_heap_region; | |
2856 } | |
2857 | |
2858 bool do_bit(size_t offset) { | |
2859 HeapWord* addr = _nextMarkBitMap->offsetToHeapWord(offset); | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
2860 assert(_nextMarkBitMap->isMarked(addr), "invariant"); |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
2861 assert( addr < _cm->finger(), "invariant"); |
342 | 2862 |
2863 if (_scanning_heap_region) { | |
2864 statsOnly( _task->increase_objs_found_on_bitmap() ); | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
2865 assert(addr >= _task->finger(), "invariant"); |
342 | 2866 // We move that task's local finger along. |
2867 _task->move_finger_to(addr); | |
2868 } else { | |
2869 // We move the task's region finger along. | |
2870 _task->move_region_finger_to(addr); | |
2871 } | |
2872 | |
2873 _task->scan_object(oop(addr)); | |
2874 // we only partially drain the local queue and global stack | |
2875 _task->drain_local_queue(true); | |
2876 _task->drain_global_stack(true); | |
2877 | |
2878 // if the has_aborted flag has been raised, we need to bail out of | |
2879 // the iteration | |
2880 return !_task->has_aborted(); | |
2881 } | |
2882 }; | |
2883 | |
2884 // Closure for iterating over objects, currently only used for | |
2885 // processing SATB buffers. | |
2886 class CMObjectClosure : public ObjectClosure { | |
2887 private: | |
2888 CMTask* _task; | |
2889 | |
2890 public: | |
2891 void do_object(oop obj) { | |
2892 _task->deal_with_reference(obj); | |
2893 } | |
2894 | |
2895 CMObjectClosure(CMTask* task) : _task(task) { } | |
2896 }; | |
2897 | |
2898 // Closure for iterating over object fields | |
2899 class CMOopClosure : public OopClosure { | |
2900 private: | |
2901 G1CollectedHeap* _g1h; | |
2902 ConcurrentMark* _cm; | |
2903 CMTask* _task; | |
2904 | |
2905 public: | |
845
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
2906 virtual void do_oop(narrowOop* p) { do_oop_work(p); } |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
2907 virtual void do_oop( oop* p) { do_oop_work(p); } |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
2908 |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
2909 template <class T> void do_oop_work(T* p) { |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
2910 assert(_g1h->is_in_g1_reserved((HeapWord*) p), "invariant"); |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
2911 assert(!_g1h->heap_region_containing((HeapWord*) p)->is_on_free_list(), |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
2912 "invariant"); |
845
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
2913 |
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
2914 oop obj = oopDesc::load_decode_heap_oop(p); |
342 | 2915 if (_cm->verbose_high()) |
2916 gclog_or_tty->print_cr("[%d] we're looking at location " | |
2917 "*"PTR_FORMAT" = "PTR_FORMAT, | |
2918 _task->task_id(), p, (void*) obj); | |
2919 _task->deal_with_reference(obj); | |
2920 } | |
2921 | |
2922 CMOopClosure(G1CollectedHeap* g1h, | |
2923 ConcurrentMark* cm, | |
2924 CMTask* task) | |
1974
fd1d227ef1b9
6983204: G1: Nightly test nsk/regression/b4958615 failing with +ExplicitGCInvokesConcurrent
johnc
parents:
1972
diff
changeset
|
2925 : _g1h(g1h), _cm(cm), _task(task) |
fd1d227ef1b9
6983204: G1: Nightly test nsk/regression/b4958615 failing with +ExplicitGCInvokesConcurrent
johnc
parents:
1972
diff
changeset
|
2926 { |
fd1d227ef1b9
6983204: G1: Nightly test nsk/regression/b4958615 failing with +ExplicitGCInvokesConcurrent
johnc
parents:
1972
diff
changeset
|
2927 _ref_processor = g1h->ref_processor(); |
fd1d227ef1b9
6983204: G1: Nightly test nsk/regression/b4958615 failing with +ExplicitGCInvokesConcurrent
johnc
parents:
1972
diff
changeset
|
2928 assert(_ref_processor != NULL, "should not be NULL"); |
fd1d227ef1b9
6983204: G1: Nightly test nsk/regression/b4958615 failing with +ExplicitGCInvokesConcurrent
johnc
parents:
1972
diff
changeset
|
2929 } |
342 | 2930 }; |
2931 | |
2932 void CMTask::setup_for_region(HeapRegion* hr) { | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
2933 // Separated the asserts so that we know which one fires. |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
2934 assert(hr != NULL, |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
2935 "claim_region() should have filtered out continues humongous regions"); |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
2936 assert(!hr->continuesHumongous(), |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
2937 "claim_region() should have filtered out continues humongous regions"); |
342 | 2938 |
2939 if (_cm->verbose_low()) | |
2940 gclog_or_tty->print_cr("[%d] setting up for region "PTR_FORMAT, | |
2941 _task_id, hr); | |
2942 | |
2943 _curr_region = hr; | |
2944 _finger = hr->bottom(); | |
2945 update_region_limit(); | |
2946 } | |
2947 | |
2948 void CMTask::update_region_limit() { | |
2949 HeapRegion* hr = _curr_region; | |
2950 HeapWord* bottom = hr->bottom(); | |
2951 HeapWord* limit = hr->next_top_at_mark_start(); | |
2952 | |
2953 if (limit == bottom) { | |
2954 if (_cm->verbose_low()) | |
2955 gclog_or_tty->print_cr("[%d] found an empty region " | |
2956 "["PTR_FORMAT", "PTR_FORMAT")", | |
2957 _task_id, bottom, limit); | |
2958 // The region was collected underneath our feet. | |
2959 // We set the finger to bottom to ensure that the bitmap | |
2960 // iteration that will follow this will not do anything. | |
2961 // (this is not a condition that holds when we set the region up, | |
2962 // as the region is not supposed to be empty in the first place) | |
2963 _finger = bottom; | |
2964 } else if (limit >= _region_limit) { | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
2965 assert(limit >= _finger, "peace of mind"); |
342 | 2966 } else { |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
2967 assert(limit < _region_limit, "only way to get here"); |
342 | 2968 // This can happen under some pretty unusual circumstances. An |
2969 // evacuation pause empties the region underneath our feet (NTAMS | |
2970 // at bottom). We then do some allocation in the region (NTAMS | |
2971 // stays at bottom), followed by the region being used as a GC | |
2972 // alloc region (NTAMS will move to top() and the objects | |
2973 // originally below it will be grayed). All objects now marked in | |
2974 // the region are explicitly grayed, if below the global finger, | |
2975 // and we do not need in fact to scan anything else. So, we simply | |
2976 // set _finger to be limit to ensure that the bitmap iteration | |
2977 // doesn't do anything. | |
2978 _finger = limit; | |
2979 } | |
2980 | |
2981 _region_limit = limit; | |
2982 } | |
2983 | |
2984 void CMTask::giveup_current_region() { | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
2985 assert(_curr_region != NULL, "invariant"); |
342 | 2986 if (_cm->verbose_low()) |
2987 gclog_or_tty->print_cr("[%d] giving up region "PTR_FORMAT, | |
2988 _task_id, _curr_region); | |
2989 clear_region_fields(); | |
2990 } | |
2991 | |
2992 void CMTask::clear_region_fields() { | |
2993 // Values for these three fields that indicate that we're not | |
2994 // holding on to a region. | |
2995 _curr_region = NULL; | |
2996 _finger = NULL; | |
2997 _region_limit = NULL; | |
2998 | |
2999 _region_finger = NULL; | |
3000 } | |
3001 | |
3002 void CMTask::reset(CMBitMap* nextMarkBitMap) { | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3003 guarantee(nextMarkBitMap != NULL, "invariant"); |
342 | 3004 |
3005 if (_cm->verbose_low()) | |
3006 gclog_or_tty->print_cr("[%d] resetting", _task_id); | |
3007 | |
3008 _nextMarkBitMap = nextMarkBitMap; | |
3009 clear_region_fields(); | |
1885
a5c514e74487
6988458: G1: assert(mr.end() <= _cm->finger()) failed: otherwise the region shouldn't be on the stack
johnc
parents:
1835
diff
changeset
|
3010 assert(_aborted_region.is_empty(), "should have been cleared"); |
342 | 3011 |
3012 _calls = 0; | |
3013 _elapsed_time_ms = 0.0; | |
3014 _termination_time_ms = 0.0; | |
3015 _termination_start_time_ms = 0.0; | |
3016 | |
3017 #if _MARKING_STATS_ | |
3018 _local_pushes = 0; | |
3019 _local_pops = 0; | |
3020 _local_max_size = 0; | |
3021 _objs_scanned = 0; | |
3022 _global_pushes = 0; | |
3023 _global_pops = 0; | |
3024 _global_max_size = 0; | |
3025 _global_transfers_to = 0; | |
3026 _global_transfers_from = 0; | |
3027 _region_stack_pops = 0; | |
3028 _regions_claimed = 0; | |
3029 _objs_found_on_bitmap = 0; | |
3030 _satb_buffers_processed = 0; | |
3031 _steal_attempts = 0; | |
3032 _steals = 0; | |
3033 _aborted = 0; | |
3034 _aborted_overflow = 0; | |
3035 _aborted_cm_aborted = 0; | |
3036 _aborted_yield = 0; | |
3037 _aborted_timed_out = 0; | |
3038 _aborted_satb = 0; | |
3039 _aborted_termination = 0; | |
3040 #endif // _MARKING_STATS_ | |
3041 } | |
3042 | |
3043 bool CMTask::should_exit_termination() { | |
3044 regular_clock_call(); | |
3045 // This is called when we are in the termination protocol. We should | |
3046 // quit if, for some reason, this task wants to abort or the global | |
3047 // stack is not empty (this means that we can get work from it). | |
3048 return !_cm->mark_stack_empty() || has_aborted(); | |
3049 } | |
3050 | |
3051 // This determines whether the method below will check both the local | |
3052 // and global fingers when determining whether to push on the stack a | |
3053 // gray object (value 1) or whether it will only check the global one | |
3054 // (value 0). The tradeoffs are that the former will be a bit more | |
3055 // accurate and possibly push less on the stack, but it might also be | |
3056 // a little bit slower. | |
3057 | |
3058 #define _CHECK_BOTH_FINGERS_ 1 | |
3059 | |
3060 void CMTask::deal_with_reference(oop obj) { | |
3061 if (_cm->verbose_high()) | |
3062 gclog_or_tty->print_cr("[%d] we're dealing with reference = "PTR_FORMAT, | |
3063 _task_id, (void*) obj); | |
3064 | |
3065 ++_refs_reached; | |
3066 | |
3067 HeapWord* objAddr = (HeapWord*) obj; | |
845
df6caf649ff7
6700789: G1: Enable use of compressed oops with G1 heaps
ysr
parents:
831
diff
changeset
|
3068 assert(obj->is_oop_or_null(true /* ignore mark word */), "Error"); |
342 | 3069 if (_g1h->is_in_g1_reserved(objAddr)) { |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3070 assert(obj != NULL, "is_in_g1_reserved should ensure this"); |
342 | 3071 HeapRegion* hr = _g1h->heap_region_containing(obj); |
3072 if (_g1h->is_obj_ill(obj, hr)) { | |
3073 if (_cm->verbose_high()) | |
3074 gclog_or_tty->print_cr("[%d] "PTR_FORMAT" is not considered marked", | |
3075 _task_id, (void*) obj); | |
3076 | |
3077 // we need to mark it first | |
3078 if (_nextMarkBitMap->parMark(objAddr)) { | |
3079 // No OrderAccess:store_load() is needed. It is implicit in the | |
3080 // CAS done in parMark(objAddr) above | |
3081 HeapWord* global_finger = _cm->finger(); | |
3082 | |
3083 #if _CHECK_BOTH_FINGERS_ | |
3084 // we will check both the local and global fingers | |
3085 | |
3086 if (_finger != NULL && objAddr < _finger) { | |
3087 if (_cm->verbose_high()) | |
3088 gclog_or_tty->print_cr("[%d] below the local finger ("PTR_FORMAT"), " | |
3089 "pushing it", _task_id, _finger); | |
3090 push(obj); | |
3091 } else if (_curr_region != NULL && objAddr < _region_limit) { | |
3092 // do nothing | |
3093 } else if (objAddr < global_finger) { | |
3094 // Notice that the global finger might be moving forward | |
3095 // concurrently. This is not a problem. In the worst case, we | |
3096 // mark the object while it is above the global finger and, by | |
3097 // the time we read the global finger, it has moved forward | |
3098 // passed this object. In this case, the object will probably | |
3099 // be visited when a task is scanning the region and will also | |
3100 // be pushed on the stack. So, some duplicate work, but no | |
3101 // correctness problems. | |
3102 | |
3103 if (_cm->verbose_high()) | |
3104 gclog_or_tty->print_cr("[%d] below the global finger " | |
3105 "("PTR_FORMAT"), pushing it", | |
3106 _task_id, global_finger); | |
3107 push(obj); | |
3108 } else { | |
3109 // do nothing | |
3110 } | |
3111 #else // _CHECK_BOTH_FINGERS_ | |
3112 // we will only check the global finger | |
3113 | |
3114 if (objAddr < global_finger) { | |
3115 // see long comment above | |
3116 | |
3117 if (_cm->verbose_high()) | |
3118 gclog_or_tty->print_cr("[%d] below the global finger " | |
3119 "("PTR_FORMAT"), pushing it", | |
3120 _task_id, global_finger); | |
3121 push(obj); | |
3122 } | |
3123 #endif // _CHECK_BOTH_FINGERS_ | |
3124 } | |
3125 } | |
3126 } | |
3127 } | |
3128 | |
3129 void CMTask::push(oop obj) { | |
3130 HeapWord* objAddr = (HeapWord*) obj; | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3131 assert(_g1h->is_in_g1_reserved(objAddr), "invariant"); |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3132 assert(!_g1h->heap_region_containing(objAddr)->is_on_free_list(), |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3133 "invariant"); |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3134 assert(!_g1h->is_obj_ill(obj), "invariant"); |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3135 assert(_nextMarkBitMap->isMarked(objAddr), "invariant"); |
342 | 3136 |
3137 if (_cm->verbose_high()) | |
3138 gclog_or_tty->print_cr("[%d] pushing "PTR_FORMAT, _task_id, (void*) obj); | |
3139 | |
3140 if (!_task_queue->push(obj)) { | |
3141 // The local task queue looks full. We need to push some entries | |
3142 // to the global stack. | |
3143 | |
3144 if (_cm->verbose_medium()) | |
3145 gclog_or_tty->print_cr("[%d] task queue overflow, " | |
3146 "moving entries to the global stack", | |
3147 _task_id); | |
3148 move_entries_to_global_stack(); | |
3149 | |
3150 // this should succeed since, even if we overflow the global | |
3151 // stack, we should have definitely removed some entries from the | |
3152 // local queue. So, there must be space on it. | |
3153 bool success = _task_queue->push(obj); | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3154 assert(success, "invariant"); |
342 | 3155 } |
3156 | |
3157 statsOnly( int tmp_size = _task_queue->size(); | |
3158 if (tmp_size > _local_max_size) | |
3159 _local_max_size = tmp_size; | |
3160 ++_local_pushes ); | |
3161 } | |
3162 | |
3163 void CMTask::reached_limit() { | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3164 assert(_words_scanned >= _words_scanned_limit || |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3165 _refs_reached >= _refs_reached_limit , |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3166 "shouldn't have been called otherwise"); |
342 | 3167 regular_clock_call(); |
3168 } | |
3169 | |
3170 void CMTask::regular_clock_call() { | |
3171 if (has_aborted()) | |
3172 return; | |
3173 | |
3174 // First, we need to recalculate the words scanned and refs reached | |
3175 // limits for the next clock call. | |
3176 recalculate_limits(); | |
3177 | |
3178 // During the regular clock call we do the following | |
3179 | |
3180 // (1) If an overflow has been flagged, then we abort. | |
3181 if (_cm->has_overflown()) { | |
3182 set_has_aborted(); | |
3183 return; | |
3184 } | |
3185 | |
3186 // If we are not concurrent (i.e. we're doing remark) we don't need | |
3187 // to check anything else. The other steps are only needed during | |
3188 // the concurrent marking phase. | |
3189 if (!concurrent()) | |
3190 return; | |
3191 | |
3192 // (2) If marking has been aborted for Full GC, then we also abort. | |
3193 if (_cm->has_aborted()) { | |
3194 set_has_aborted(); | |
3195 statsOnly( ++_aborted_cm_aborted ); | |
3196 return; | |
3197 } | |
3198 | |
3199 double curr_time_ms = os::elapsedVTime() * 1000.0; | |
3200 | |
3201 // (3) If marking stats are enabled, then we update the step history. | |
3202 #if _MARKING_STATS_ | |
3203 if (_words_scanned >= _words_scanned_limit) | |
3204 ++_clock_due_to_scanning; | |
3205 if (_refs_reached >= _refs_reached_limit) | |
3206 ++_clock_due_to_marking; | |
3207 | |
3208 double last_interval_ms = curr_time_ms - _interval_start_time_ms; | |
3209 _interval_start_time_ms = curr_time_ms; | |
3210 _all_clock_intervals_ms.add(last_interval_ms); | |
3211 | |
3212 if (_cm->verbose_medium()) { | |
3213 gclog_or_tty->print_cr("[%d] regular clock, interval = %1.2lfms, " | |
3214 "scanned = %d%s, refs reached = %d%s", | |
3215 _task_id, last_interval_ms, | |
3216 _words_scanned, | |
3217 (_words_scanned >= _words_scanned_limit) ? " (*)" : "", | |
3218 _refs_reached, | |
3219 (_refs_reached >= _refs_reached_limit) ? " (*)" : ""); | |
3220 } | |
3221 #endif // _MARKING_STATS_ | |
3222 | |
3223 // (4) We check whether we should yield. If we have to, then we abort. | |
3224 if (_cm->should_yield()) { | |
3225 // We should yield. To do this we abort the task. The caller is | |
3226 // responsible for yielding. | |
3227 set_has_aborted(); | |
3228 statsOnly( ++_aborted_yield ); | |
3229 return; | |
3230 } | |
3231 | |
3232 // (5) We check whether we've reached our time quota. If we have, | |
3233 // then we abort. | |
3234 double elapsed_time_ms = curr_time_ms - _start_time_ms; | |
3235 if (elapsed_time_ms > _time_target_ms) { | |
3236 set_has_aborted(); | |
3237 _has_aborted_timed_out = true; | |
3238 statsOnly( ++_aborted_timed_out ); | |
3239 return; | |
3240 } | |
3241 | |
3242 // (6) Finally, we check whether there are enough completed STAB | |
3243 // buffers available for processing. If there are, we abort. | |
3244 SATBMarkQueueSet& satb_mq_set = JavaThread::satb_mark_queue_set(); | |
3245 if (!_draining_satb_buffers && satb_mq_set.process_completed_buffers()) { | |
3246 if (_cm->verbose_low()) | |
3247 gclog_or_tty->print_cr("[%d] aborting to deal with pending SATB buffers", | |
3248 _task_id); | |
3249 // we do need to process SATB buffers, we'll abort and restart | |
3250 // the marking task to do so | |
3251 set_has_aborted(); | |
3252 statsOnly( ++_aborted_satb ); | |
3253 return; | |
3254 } | |
3255 } | |
3256 | |
3257 void CMTask::recalculate_limits() { | |
3258 _real_words_scanned_limit = _words_scanned + words_scanned_period; | |
3259 _words_scanned_limit = _real_words_scanned_limit; | |
3260 | |
3261 _real_refs_reached_limit = _refs_reached + refs_reached_period; | |
3262 _refs_reached_limit = _real_refs_reached_limit; | |
3263 } | |
3264 | |
3265 void CMTask::decrease_limits() { | |
3266 // This is called when we believe that we're going to do an infrequent | |
3267 // operation which will increase the per byte scanned cost (i.e. move | |
3268 // entries to/from the global stack). It basically tries to decrease the | |
3269 // scanning limit so that the clock is called earlier. | |
3270 | |
3271 if (_cm->verbose_medium()) | |
3272 gclog_or_tty->print_cr("[%d] decreasing limits", _task_id); | |
3273 | |
3274 _words_scanned_limit = _real_words_scanned_limit - | |
3275 3 * words_scanned_period / 4; | |
3276 _refs_reached_limit = _real_refs_reached_limit - | |
3277 3 * refs_reached_period / 4; | |
3278 } | |
3279 | |
3280 void CMTask::move_entries_to_global_stack() { | |
3281 // local array where we'll store the entries that will be popped | |
3282 // from the local queue | |
3283 oop buffer[global_stack_transfer_size]; | |
3284 | |
3285 int n = 0; | |
3286 oop obj; | |
3287 while (n < global_stack_transfer_size && _task_queue->pop_local(obj)) { | |
3288 buffer[n] = obj; | |
3289 ++n; | |
3290 } | |
3291 | |
3292 if (n > 0) { | |
3293 // we popped at least one entry from the local queue | |
3294 | |
3295 statsOnly( ++_global_transfers_to; _local_pops += n ); | |
3296 | |
3297 if (!_cm->mark_stack_push(buffer, n)) { | |
3298 if (_cm->verbose_low()) | |
3299 gclog_or_tty->print_cr("[%d] aborting due to global stack overflow", _task_id); | |
3300 set_has_aborted(); | |
3301 } else { | |
3302 // the transfer was successful | |
3303 | |
3304 if (_cm->verbose_medium()) | |
3305 gclog_or_tty->print_cr("[%d] pushed %d entries to the global stack", | |
3306 _task_id, n); | |
3307 statsOnly( int tmp_size = _cm->mark_stack_size(); | |
3308 if (tmp_size > _global_max_size) | |
3309 _global_max_size = tmp_size; | |
3310 _global_pushes += n ); | |
3311 } | |
3312 } | |
3313 | |
3314 // this operation was quite expensive, so decrease the limits | |
3315 decrease_limits(); | |
3316 } | |
3317 | |
3318 void CMTask::get_entries_from_global_stack() { | |
3319 // local array where we'll store the entries that will be popped | |
3320 // from the global stack. | |
3321 oop buffer[global_stack_transfer_size]; | |
3322 int n; | |
3323 _cm->mark_stack_pop(buffer, global_stack_transfer_size, &n); | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3324 assert(n <= global_stack_transfer_size, |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3325 "we should not pop more than the given limit"); |
342 | 3326 if (n > 0) { |
3327 // yes, we did actually pop at least one entry | |
3328 | |
3329 statsOnly( ++_global_transfers_from; _global_pops += n ); | |
3330 if (_cm->verbose_medium()) | |
3331 gclog_or_tty->print_cr("[%d] popped %d entries from the global stack", | |
3332 _task_id, n); | |
3333 for (int i = 0; i < n; ++i) { | |
3334 bool success = _task_queue->push(buffer[i]); | |
3335 // We only call this when the local queue is empty or under a | |
3336 // given target limit. So, we do not expect this push to fail. | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3337 assert(success, "invariant"); |
342 | 3338 } |
3339 | |
3340 statsOnly( int tmp_size = _task_queue->size(); | |
3341 if (tmp_size > _local_max_size) | |
3342 _local_max_size = tmp_size; | |
3343 _local_pushes += n ); | |
3344 } | |
3345 | |
3346 // this operation was quite expensive, so decrease the limits | |
3347 decrease_limits(); | |
3348 } | |
3349 | |
3350 void CMTask::drain_local_queue(bool partially) { | |
3351 if (has_aborted()) | |
3352 return; | |
3353 | |
3354 // Decide what the target size is, depending whether we're going to | |
3355 // drain it partially (so that other tasks can steal if they run out | |
3356 // of things to do) or totally (at the very end). | |
3357 size_t target_size; | |
3358 if (partially) | |
3359 target_size = MIN2((size_t)_task_queue->max_elems()/3, GCDrainStackTargetSize); | |
3360 else | |
3361 target_size = 0; | |
3362 | |
3363 if (_task_queue->size() > target_size) { | |
3364 if (_cm->verbose_high()) | |
3365 gclog_or_tty->print_cr("[%d] draining local queue, target size = %d", | |
3366 _task_id, target_size); | |
3367 | |
3368 oop obj; | |
3369 bool ret = _task_queue->pop_local(obj); | |
3370 while (ret) { | |
3371 statsOnly( ++_local_pops ); | |
3372 | |
3373 if (_cm->verbose_high()) | |
3374 gclog_or_tty->print_cr("[%d] popped "PTR_FORMAT, _task_id, | |
3375 (void*) obj); | |
3376 | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3377 assert(_g1h->is_in_g1_reserved((HeapWord*) obj), "invariant" ); |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3378 assert(!_g1h->heap_region_containing(obj)->is_on_free_list(), |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3379 "invariant"); |
342 | 3380 |
3381 scan_object(obj); | |
3382 | |
3383 if (_task_queue->size() <= target_size || has_aborted()) | |
3384 ret = false; | |
3385 else | |
3386 ret = _task_queue->pop_local(obj); | |
3387 } | |
3388 | |
3389 if (_cm->verbose_high()) | |
3390 gclog_or_tty->print_cr("[%d] drained local queue, size = %d", | |
3391 _task_id, _task_queue->size()); | |
3392 } | |
3393 } | |
3394 | |
3395 void CMTask::drain_global_stack(bool partially) { | |
3396 if (has_aborted()) | |
3397 return; | |
3398 | |
3399 // We have a policy to drain the local queue before we attempt to | |
3400 // drain the global stack. | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3401 assert(partially || _task_queue->size() == 0, "invariant"); |
342 | 3402 |
3403 // Decide what the target size is, depending whether we're going to | |
3404 // drain it partially (so that other tasks can steal if they run out | |
3405 // of things to do) or totally (at the very end). Notice that, | |
3406 // because we move entries from the global stack in chunks or | |
3407 // because another task might be doing the same, we might in fact | |
3408 // drop below the target. But, this is not a problem. | |
3409 size_t target_size; | |
3410 if (partially) | |
3411 target_size = _cm->partial_mark_stack_size_target(); | |
3412 else | |
3413 target_size = 0; | |
3414 | |
3415 if (_cm->mark_stack_size() > target_size) { | |
3416 if (_cm->verbose_low()) | |
3417 gclog_or_tty->print_cr("[%d] draining global_stack, target size %d", | |
3418 _task_id, target_size); | |
3419 | |
3420 while (!has_aborted() && _cm->mark_stack_size() > target_size) { | |
3421 get_entries_from_global_stack(); | |
3422 drain_local_queue(partially); | |
3423 } | |
3424 | |
3425 if (_cm->verbose_low()) | |
3426 gclog_or_tty->print_cr("[%d] drained global stack, size = %d", | |
3427 _task_id, _cm->mark_stack_size()); | |
3428 } | |
3429 } | |
3430 | |
3431 // SATB Queue has several assumptions on whether to call the par or | |
3432 // non-par versions of the methods. this is why some of the code is | |
3433 // replicated. We should really get rid of the single-threaded version | |
3434 // of the code to simplify things. | |
3435 void CMTask::drain_satb_buffers() { | |
3436 if (has_aborted()) | |
3437 return; | |
3438 | |
3439 // We set this so that the regular clock knows that we're in the | |
3440 // middle of draining buffers and doesn't set the abort flag when it | |
3441 // notices that SATB buffers are available for draining. It'd be | |
3442 // very counter productive if it did that. :-) | |
3443 _draining_satb_buffers = true; | |
3444 | |
3445 CMObjectClosure oc(this); | |
3446 SATBMarkQueueSet& satb_mq_set = JavaThread::satb_mark_queue_set(); | |
1833
8b10f48633dc
6984287: Regularize how GC parallel workers are specified.
jmasa
parents:
1719
diff
changeset
|
3447 if (G1CollectedHeap::use_parallel_gc_threads()) |
342 | 3448 satb_mq_set.set_par_closure(_task_id, &oc); |
3449 else | |
3450 satb_mq_set.set_closure(&oc); | |
3451 | |
3452 // This keeps claiming and applying the closure to completed buffers | |
3453 // until we run out of buffers or we need to abort. | |
1833
8b10f48633dc
6984287: Regularize how GC parallel workers are specified.
jmasa
parents:
1719
diff
changeset
|
3454 if (G1CollectedHeap::use_parallel_gc_threads()) { |
342 | 3455 while (!has_aborted() && |
3456 satb_mq_set.par_apply_closure_to_completed_buffer(_task_id)) { | |
3457 if (_cm->verbose_medium()) | |
3458 gclog_or_tty->print_cr("[%d] processed an SATB buffer", _task_id); | |
3459 statsOnly( ++_satb_buffers_processed ); | |
3460 regular_clock_call(); | |
3461 } | |
3462 } else { | |
3463 while (!has_aborted() && | |
3464 satb_mq_set.apply_closure_to_completed_buffer()) { | |
3465 if (_cm->verbose_medium()) | |
3466 gclog_or_tty->print_cr("[%d] processed an SATB buffer", _task_id); | |
3467 statsOnly( ++_satb_buffers_processed ); | |
3468 regular_clock_call(); | |
3469 } | |
3470 } | |
3471 | |
3472 if (!concurrent() && !has_aborted()) { | |
3473 // We should only do this during remark. | |
1833
8b10f48633dc
6984287: Regularize how GC parallel workers are specified.
jmasa
parents:
1719
diff
changeset
|
3474 if (G1CollectedHeap::use_parallel_gc_threads()) |
342 | 3475 satb_mq_set.par_iterate_closure_all_threads(_task_id); |
3476 else | |
3477 satb_mq_set.iterate_closure_all_threads(); | |
3478 } | |
3479 | |
3480 _draining_satb_buffers = false; | |
3481 | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3482 assert(has_aborted() || |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3483 concurrent() || |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3484 satb_mq_set.completed_buffers_num() == 0, "invariant"); |
342 | 3485 |
1833
8b10f48633dc
6984287: Regularize how GC parallel workers are specified.
jmasa
parents:
1719
diff
changeset
|
3486 if (G1CollectedHeap::use_parallel_gc_threads()) |
342 | 3487 satb_mq_set.set_par_closure(_task_id, NULL); |
3488 else | |
3489 satb_mq_set.set_closure(NULL); | |
3490 | |
3491 // again, this was a potentially expensive operation, decrease the | |
3492 // limits to get the regular clock call early | |
3493 decrease_limits(); | |
3494 } | |
3495 | |
3496 void CMTask::drain_region_stack(BitMapClosure* bc) { | |
3497 if (has_aborted()) | |
3498 return; | |
3499 | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3500 assert(_region_finger == NULL, |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3501 "it should be NULL when we're not scanning a region"); |
342 | 3502 |
1835
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
3503 if (!_cm->region_stack_empty() || !_aborted_region.is_empty()) { |
342 | 3504 if (_cm->verbose_low()) |
3505 gclog_or_tty->print_cr("[%d] draining region stack, size = %d", | |
3506 _task_id, _cm->region_stack_size()); | |
3507 | |
1835
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
3508 MemRegion mr; |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
3509 |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
3510 if (!_aborted_region.is_empty()) { |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
3511 mr = _aborted_region; |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
3512 _aborted_region = MemRegion(); |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
3513 |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
3514 if (_cm->verbose_low()) |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
3515 gclog_or_tty->print_cr("[%d] scanning aborted region [ " PTR_FORMAT ", " PTR_FORMAT " )", |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
3516 _task_id, mr.start(), mr.end()); |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
3517 } else { |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
3518 mr = _cm->region_stack_pop_lock_free(); |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
3519 // it returns MemRegion() if the pop fails |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
3520 statsOnly(if (mr.start() != NULL) ++_region_stack_pops ); |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
3521 } |
342 | 3522 |
3523 while (mr.start() != NULL) { | |
3524 if (_cm->verbose_medium()) | |
3525 gclog_or_tty->print_cr("[%d] we are scanning region " | |
3526 "["PTR_FORMAT", "PTR_FORMAT")", | |
3527 _task_id, mr.start(), mr.end()); | |
1835
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
3528 |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3529 assert(mr.end() <= _cm->finger(), |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3530 "otherwise the region shouldn't be on the stack"); |
342 | 3531 assert(!mr.is_empty(), "Only non-empty regions live on the region stack"); |
3532 if (_nextMarkBitMap->iterate(bc, mr)) { | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3533 assert(!has_aborted(), |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3534 "cannot abort the task without aborting the bitmap iteration"); |
342 | 3535 |
3536 // We finished iterating over the region without aborting. | |
3537 regular_clock_call(); | |
3538 if (has_aborted()) | |
3539 mr = MemRegion(); | |
3540 else { | |
1835
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
3541 mr = _cm->region_stack_pop_lock_free(); |
342 | 3542 // it returns MemRegion() if the pop fails |
3543 statsOnly(if (mr.start() != NULL) ++_region_stack_pops ); | |
3544 } | |
3545 } else { | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3546 assert(has_aborted(), "currently the only way to do so"); |
342 | 3547 |
3548 // The only way to abort the bitmap iteration is to return | |
3549 // false from the do_bit() method. However, inside the | |
3550 // do_bit() method we move the _region_finger to point to the | |
3551 // object currently being looked at. So, if we bail out, we | |
3552 // have definitely set _region_finger to something non-null. | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3553 assert(_region_finger != NULL, "invariant"); |
342 | 3554 |
1835
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
3555 // Make sure that any previously aborted region has been |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
3556 // cleared. |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
3557 assert(_aborted_region.is_empty(), "aborted region not cleared"); |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
3558 |
342 | 3559 // The iteration was actually aborted. So now _region_finger |
3560 // points to the address of the object we last scanned. If we | |
3561 // leave it there, when we restart this task, we will rescan | |
3562 // the object. It is easy to avoid this. We move the finger by | |
3563 // enough to point to the next possible object header (the | |
3564 // bitmap knows by how much we need to move it as it knows its | |
3565 // granularity). | |
3566 MemRegion newRegion = | |
3567 MemRegion(_nextMarkBitMap->nextWord(_region_finger), mr.end()); | |
3568 | |
3569 if (!newRegion.is_empty()) { | |
3570 if (_cm->verbose_low()) { | |
1835
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
3571 gclog_or_tty->print_cr("[%d] recording unscanned region" |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
3572 "[" PTR_FORMAT "," PTR_FORMAT ") in CMTask", |
342 | 3573 _task_id, |
3574 newRegion.start(), newRegion.end()); | |
3575 } | |
1835
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
3576 // Now record the part of the region we didn't scan to |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
3577 // make sure this task scans it later. |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
3578 _aborted_region = newRegion; |
342 | 3579 } |
3580 // break from while | |
3581 mr = MemRegion(); | |
3582 } | |
3583 _region_finger = NULL; | |
3584 } | |
3585 | |
3586 if (_cm->verbose_low()) | |
3587 gclog_or_tty->print_cr("[%d] drained region stack, size = %d", | |
3588 _task_id, _cm->region_stack_size()); | |
3589 } | |
3590 } | |
3591 | |
3592 void CMTask::print_stats() { | |
3593 gclog_or_tty->print_cr("Marking Stats, task = %d, calls = %d", | |
3594 _task_id, _calls); | |
3595 gclog_or_tty->print_cr(" Elapsed time = %1.2lfms, Termination time = %1.2lfms", | |
3596 _elapsed_time_ms, _termination_time_ms); | |
3597 gclog_or_tty->print_cr(" Step Times (cum): num = %d, avg = %1.2lfms, sd = %1.2lfms", | |
3598 _step_times_ms.num(), _step_times_ms.avg(), | |
3599 _step_times_ms.sd()); | |
3600 gclog_or_tty->print_cr(" max = %1.2lfms, total = %1.2lfms", | |
3601 _step_times_ms.maximum(), _step_times_ms.sum()); | |
3602 | |
3603 #if _MARKING_STATS_ | |
3604 gclog_or_tty->print_cr(" Clock Intervals (cum): num = %d, avg = %1.2lfms, sd = %1.2lfms", | |
3605 _all_clock_intervals_ms.num(), _all_clock_intervals_ms.avg(), | |
3606 _all_clock_intervals_ms.sd()); | |
3607 gclog_or_tty->print_cr(" max = %1.2lfms, total = %1.2lfms", | |
3608 _all_clock_intervals_ms.maximum(), | |
3609 _all_clock_intervals_ms.sum()); | |
3610 gclog_or_tty->print_cr(" Clock Causes (cum): scanning = %d, marking = %d", | |
3611 _clock_due_to_scanning, _clock_due_to_marking); | |
3612 gclog_or_tty->print_cr(" Objects: scanned = %d, found on the bitmap = %d", | |
3613 _objs_scanned, _objs_found_on_bitmap); | |
3614 gclog_or_tty->print_cr(" Local Queue: pushes = %d, pops = %d, max size = %d", | |
3615 _local_pushes, _local_pops, _local_max_size); | |
3616 gclog_or_tty->print_cr(" Global Stack: pushes = %d, pops = %d, max size = %d", | |
3617 _global_pushes, _global_pops, _global_max_size); | |
3618 gclog_or_tty->print_cr(" transfers to = %d, transfers from = %d", | |
3619 _global_transfers_to,_global_transfers_from); | |
3620 gclog_or_tty->print_cr(" Regions: claimed = %d, Region Stack: pops = %d", | |
3621 _regions_claimed, _region_stack_pops); | |
3622 gclog_or_tty->print_cr(" SATB buffers: processed = %d", _satb_buffers_processed); | |
3623 gclog_or_tty->print_cr(" Steals: attempts = %d, successes = %d", | |
3624 _steal_attempts, _steals); | |
3625 gclog_or_tty->print_cr(" Aborted: %d, due to", _aborted); | |
3626 gclog_or_tty->print_cr(" overflow: %d, global abort: %d, yield: %d", | |
3627 _aborted_overflow, _aborted_cm_aborted, _aborted_yield); | |
3628 gclog_or_tty->print_cr(" time out: %d, SATB: %d, termination: %d", | |
3629 _aborted_timed_out, _aborted_satb, _aborted_termination); | |
3630 #endif // _MARKING_STATS_ | |
3631 } | |
3632 | |
3633 /***************************************************************************** | |
3634 | |
3635 The do_marking_step(time_target_ms) method is the building block | |
3636 of the parallel marking framework. It can be called in parallel | |
3637 with other invocations of do_marking_step() on different tasks | |
3638 (but only one per task, obviously) and concurrently with the | |
3639 mutator threads, or during remark, hence it eliminates the need | |
3640 for two versions of the code. When called during remark, it will | |
3641 pick up from where the task left off during the concurrent marking | |
3642 phase. Interestingly, tasks are also claimable during evacuation | |
3643 pauses too, since do_marking_step() ensures that it aborts before | |
3644 it needs to yield. | |
3645 | |
3646 The data structures that is uses to do marking work are the | |
3647 following: | |
3648 | |
3649 (1) Marking Bitmap. If there are gray objects that appear only | |
3650 on the bitmap (this happens either when dealing with an overflow | |
3651 or when the initial marking phase has simply marked the roots | |
3652 and didn't push them on the stack), then tasks claim heap | |
3653 regions whose bitmap they then scan to find gray objects. A | |
3654 global finger indicates where the end of the last claimed region | |
3655 is. A local finger indicates how far into the region a task has | |
3656 scanned. The two fingers are used to determine how to gray an | |
3657 object (i.e. whether simply marking it is OK, as it will be | |
3658 visited by a task in the future, or whether it needs to be also | |
3659 pushed on a stack). | |
3660 | |
3661 (2) Local Queue. The local queue of the task which is accessed | |
3662 reasonably efficiently by the task. Other tasks can steal from | |
3663 it when they run out of work. Throughout the marking phase, a | |
3664 task attempts to keep its local queue short but not totally | |
3665 empty, so that entries are available for stealing by other | |
3666 tasks. Only when there is no more work, a task will totally | |
3667 drain its local queue. | |
3668 | |
3669 (3) Global Mark Stack. This handles local queue overflow. During | |
3670 marking only sets of entries are moved between it and the local | |
3671 queues, as access to it requires a mutex and more fine-grain | |
3672 interaction with it which might cause contention. If it | |
3673 overflows, then the marking phase should restart and iterate | |
3674 over the bitmap to identify gray objects. Throughout the marking | |
3675 phase, tasks attempt to keep the global mark stack at a small | |
3676 length but not totally empty, so that entries are available for | |
3677 popping by other tasks. Only when there is no more work, tasks | |
3678 will totally drain the global mark stack. | |
3679 | |
3680 (4) Global Region Stack. Entries on it correspond to areas of | |
3681 the bitmap that need to be scanned since they contain gray | |
3682 objects. Pushes on the region stack only happen during | |
3683 evacuation pauses and typically correspond to areas covered by | |
3684 GC LABS. If it overflows, then the marking phase should restart | |
3685 and iterate over the bitmap to identify gray objects. Tasks will | |
3686 try to totally drain the region stack as soon as possible. | |
3687 | |
3688 (5) SATB Buffer Queue. This is where completed SATB buffers are | |
3689 made available. Buffers are regularly removed from this queue | |
3690 and scanned for roots, so that the queue doesn't get too | |
3691 long. During remark, all completed buffers are processed, as | |
3692 well as the filled in parts of any uncompleted buffers. | |
3693 | |
3694 The do_marking_step() method tries to abort when the time target | |
3695 has been reached. There are a few other cases when the | |
3696 do_marking_step() method also aborts: | |
3697 | |
3698 (1) When the marking phase has been aborted (after a Full GC). | |
3699 | |
3700 (2) When a global overflow (either on the global stack or the | |
3701 region stack) has been triggered. Before the task aborts, it | |
3702 will actually sync up with the other tasks to ensure that all | |
3703 the marking data structures (local queues, stacks, fingers etc.) | |
3704 are re-initialised so that when do_marking_step() completes, | |
3705 the marking phase can immediately restart. | |
3706 | |
3707 (3) When enough completed SATB buffers are available. The | |
3708 do_marking_step() method only tries to drain SATB buffers right | |
3709 at the beginning. So, if enough buffers are available, the | |
3710 marking step aborts and the SATB buffers are processed at | |
3711 the beginning of the next invocation. | |
3712 | |
3713 (4) To yield. when we have to yield then we abort and yield | |
3714 right at the end of do_marking_step(). This saves us from a lot | |
3715 of hassle as, by yielding we might allow a Full GC. If this | |
3716 happens then objects will be compacted underneath our feet, the | |
3717 heap might shrink, etc. We save checking for this by just | |
3718 aborting and doing the yield right at the end. | |
3719 | |
3720 From the above it follows that the do_marking_step() method should | |
3721 be called in a loop (or, otherwise, regularly) until it completes. | |
3722 | |
3723 If a marking step completes without its has_aborted() flag being | |
3724 true, it means it has completed the current marking phase (and | |
3725 also all other marking tasks have done so and have all synced up). | |
3726 | |
3727 A method called regular_clock_call() is invoked "regularly" (in | |
3728 sub ms intervals) throughout marking. It is this clock method that | |
3729 checks all the abort conditions which were mentioned above and | |
3730 decides when the task should abort. A work-based scheme is used to | |
3731 trigger this clock method: when the number of object words the | |
3732 marking phase has scanned or the number of references the marking | |
3733 phase has visited reach a given limit. Additional invocations to | |
3734 the method clock have been planted in a few other strategic places | |
3735 too. The initial reason for the clock method was to avoid calling | |
3736 vtime too regularly, as it is quite expensive. So, once it was in | |
3737 place, it was natural to piggy-back all the other conditions on it | |
3738 too and not constantly check them throughout the code. | |
3739 | |
3740 *****************************************************************************/ | |
3741 | |
3742 void CMTask::do_marking_step(double time_target_ms) { | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3743 assert(time_target_ms >= 1.0, "minimum granularity is 1ms"); |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3744 assert(concurrent() == _cm->concurrent(), "they should be the same"); |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3745 |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3746 assert(concurrent() || _cm->region_stack_empty(), |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3747 "the region stack should have been cleared before remark"); |
1835
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
3748 assert(concurrent() || !_cm->has_aborted_regions(), |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
3749 "aborted regions should have been cleared before remark"); |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3750 assert(_region_finger == NULL, |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3751 "this should be non-null only when a region is being scanned"); |
342 | 3752 |
3753 G1CollectorPolicy* g1_policy = _g1h->g1_policy(); | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3754 assert(_task_queues != NULL, "invariant"); |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3755 assert(_task_queue != NULL, "invariant"); |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3756 assert(_task_queues->queue(_task_id) == _task_queue, "invariant"); |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3757 |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3758 assert(!_claimed, |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3759 "only one thread should claim this task at any one time"); |
342 | 3760 |
3761 // OK, this doesn't safeguard again all possible scenarios, as it is | |
3762 // possible for two threads to set the _claimed flag at the same | |
3763 // time. But it is only for debugging purposes anyway and it will | |
3764 // catch most problems. | |
3765 _claimed = true; | |
3766 | |
3767 _start_time_ms = os::elapsedVTime() * 1000.0; | |
3768 statsOnly( _interval_start_time_ms = _start_time_ms ); | |
3769 | |
3770 double diff_prediction_ms = | |
3771 g1_policy->get_new_prediction(&_marking_step_diffs_ms); | |
3772 _time_target_ms = time_target_ms - diff_prediction_ms; | |
3773 | |
3774 // set up the variables that are used in the work-based scheme to | |
3775 // call the regular clock method | |
3776 _words_scanned = 0; | |
3777 _refs_reached = 0; | |
3778 recalculate_limits(); | |
3779 | |
3780 // clear all flags | |
3781 clear_has_aborted(); | |
3782 _has_aborted_timed_out = false; | |
3783 _draining_satb_buffers = false; | |
3784 | |
3785 ++_calls; | |
3786 | |
3787 if (_cm->verbose_low()) | |
3788 gclog_or_tty->print_cr("[%d] >>>>>>>>>> START, call = %d, " | |
3789 "target = %1.2lfms >>>>>>>>>>", | |
3790 _task_id, _calls, _time_target_ms); | |
3791 | |
3792 // Set up the bitmap and oop closures. Anything that uses them is | |
3793 // eventually called from this method, so it is OK to allocate these | |
3794 // statically. | |
3795 CMBitMapClosure bitmap_closure(this, _cm, _nextMarkBitMap); | |
3796 CMOopClosure oop_closure(_g1h, _cm, this); | |
3797 set_oop_closure(&oop_closure); | |
3798 | |
3799 if (_cm->has_overflown()) { | |
3800 // This can happen if the region stack or the mark stack overflows | |
3801 // during a GC pause and this task, after a yield point, | |
3802 // restarts. We have to abort as we need to get into the overflow | |
3803 // protocol which happens right at the end of this task. | |
3804 set_has_aborted(); | |
3805 } | |
3806 | |
3807 // First drain any available SATB buffers. After this, we will not | |
3808 // look at SATB buffers before the next invocation of this method. | |
3809 // If enough completed SATB buffers are queued up, the regular clock | |
3810 // will abort this task so that it restarts. | |
3811 drain_satb_buffers(); | |
3812 // ...then partially drain the local queue and the global stack | |
3813 drain_local_queue(true); | |
3814 drain_global_stack(true); | |
3815 | |
3816 // Then totally drain the region stack. We will not look at | |
3817 // it again before the next invocation of this method. Entries on | |
3818 // the region stack are only added during evacuation pauses, for | |
3819 // which we have to yield. When we do, we abort the task anyway so | |
3820 // it will look at the region stack again when it restarts. | |
3821 bitmap_closure.set_scanning_heap_region(false); | |
3822 drain_region_stack(&bitmap_closure); | |
3823 // ...then partially drain the local queue and the global stack | |
3824 drain_local_queue(true); | |
3825 drain_global_stack(true); | |
3826 | |
3827 do { | |
3828 if (!has_aborted() && _curr_region != NULL) { | |
3829 // This means that we're already holding on to a region. | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3830 assert(_finger != NULL, "if region is not NULL, then the finger " |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3831 "should not be NULL either"); |
342 | 3832 |
3833 // We might have restarted this task after an evacuation pause | |
3834 // which might have evacuated the region we're holding on to | |
3835 // underneath our feet. Let's read its limit again to make sure | |
3836 // that we do not iterate over a region of the heap that | |
3837 // contains garbage (update_region_limit() will also move | |
3838 // _finger to the start of the region if it is found empty). | |
3839 update_region_limit(); | |
3840 // We will start from _finger not from the start of the region, | |
3841 // as we might be restarting this task after aborting half-way | |
3842 // through scanning this region. In this case, _finger points to | |
3843 // the address where we last found a marked object. If this is a | |
3844 // fresh region, _finger points to start(). | |
3845 MemRegion mr = MemRegion(_finger, _region_limit); | |
3846 | |
3847 if (_cm->verbose_low()) | |
3848 gclog_or_tty->print_cr("[%d] we're scanning part " | |
3849 "["PTR_FORMAT", "PTR_FORMAT") " | |
3850 "of region "PTR_FORMAT, | |
3851 _task_id, _finger, _region_limit, _curr_region); | |
3852 | |
3853 // Let's iterate over the bitmap of the part of the | |
3854 // region that is left. | |
3855 bitmap_closure.set_scanning_heap_region(true); | |
3856 if (mr.is_empty() || | |
3857 _nextMarkBitMap->iterate(&bitmap_closure, mr)) { | |
3858 // We successfully completed iterating over the region. Now, | |
3859 // let's give up the region. | |
3860 giveup_current_region(); | |
3861 regular_clock_call(); | |
3862 } else { | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3863 assert(has_aborted(), "currently the only way to do so"); |
342 | 3864 // The only way to abort the bitmap iteration is to return |
3865 // false from the do_bit() method. However, inside the | |
3866 // do_bit() method we move the _finger to point to the | |
3867 // object currently being looked at. So, if we bail out, we | |
3868 // have definitely set _finger to something non-null. | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3869 assert(_finger != NULL, "invariant"); |
342 | 3870 |
3871 // Region iteration was actually aborted. So now _finger | |
3872 // points to the address of the object we last scanned. If we | |
3873 // leave it there, when we restart this task, we will rescan | |
3874 // the object. It is easy to avoid this. We move the finger by | |
3875 // enough to point to the next possible object header (the | |
3876 // bitmap knows by how much we need to move it as it knows its | |
3877 // granularity). | |
1314
3f0549ed0c98
6921710: G1: assert(new_finger >= _finger && new_finger < _region_limit,"invariant")
apetrusenko
parents:
1284
diff
changeset
|
3878 assert(_finger < _region_limit, "invariant"); |
3f0549ed0c98
6921710: G1: assert(new_finger >= _finger && new_finger < _region_limit,"invariant")
apetrusenko
parents:
1284
diff
changeset
|
3879 HeapWord* new_finger = _nextMarkBitMap->nextWord(_finger); |
3f0549ed0c98
6921710: G1: assert(new_finger >= _finger && new_finger < _region_limit,"invariant")
apetrusenko
parents:
1284
diff
changeset
|
3880 // Check if bitmap iteration was aborted while scanning the last object |
3f0549ed0c98
6921710: G1: assert(new_finger >= _finger && new_finger < _region_limit,"invariant")
apetrusenko
parents:
1284
diff
changeset
|
3881 if (new_finger >= _region_limit) { |
3f0549ed0c98
6921710: G1: assert(new_finger >= _finger && new_finger < _region_limit,"invariant")
apetrusenko
parents:
1284
diff
changeset
|
3882 giveup_current_region(); |
3f0549ed0c98
6921710: G1: assert(new_finger >= _finger && new_finger < _region_limit,"invariant")
apetrusenko
parents:
1284
diff
changeset
|
3883 } else { |
3f0549ed0c98
6921710: G1: assert(new_finger >= _finger && new_finger < _region_limit,"invariant")
apetrusenko
parents:
1284
diff
changeset
|
3884 move_finger_to(new_finger); |
3f0549ed0c98
6921710: G1: assert(new_finger >= _finger && new_finger < _region_limit,"invariant")
apetrusenko
parents:
1284
diff
changeset
|
3885 } |
342 | 3886 } |
3887 } | |
3888 // At this point we have either completed iterating over the | |
3889 // region we were holding on to, or we have aborted. | |
3890 | |
3891 // We then partially drain the local queue and the global stack. | |
3892 // (Do we really need this?) | |
3893 drain_local_queue(true); | |
3894 drain_global_stack(true); | |
3895 | |
3896 // Read the note on the claim_region() method on why it might | |
3897 // return NULL with potentially more regions available for | |
3898 // claiming and why we have to check out_of_regions() to determine | |
3899 // whether we're done or not. | |
3900 while (!has_aborted() && _curr_region == NULL && !_cm->out_of_regions()) { | |
3901 // We are going to try to claim a new region. We should have | |
3902 // given up on the previous one. | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3903 // Separated the asserts so that we know which one fires. |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3904 assert(_curr_region == NULL, "invariant"); |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3905 assert(_finger == NULL, "invariant"); |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3906 assert(_region_limit == NULL, "invariant"); |
342 | 3907 if (_cm->verbose_low()) |
3908 gclog_or_tty->print_cr("[%d] trying to claim a new region", _task_id); | |
3909 HeapRegion* claimed_region = _cm->claim_region(_task_id); | |
3910 if (claimed_region != NULL) { | |
3911 // Yes, we managed to claim one | |
3912 statsOnly( ++_regions_claimed ); | |
3913 | |
3914 if (_cm->verbose_low()) | |
3915 gclog_or_tty->print_cr("[%d] we successfully claimed " | |
3916 "region "PTR_FORMAT, | |
3917 _task_id, claimed_region); | |
3918 | |
3919 setup_for_region(claimed_region); | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3920 assert(_curr_region == claimed_region, "invariant"); |
342 | 3921 } |
3922 // It is important to call the regular clock here. It might take | |
3923 // a while to claim a region if, for example, we hit a large | |
3924 // block of empty regions. So we need to call the regular clock | |
3925 // method once round the loop to make sure it's called | |
3926 // frequently enough. | |
3927 regular_clock_call(); | |
3928 } | |
3929 | |
3930 if (!has_aborted() && _curr_region == NULL) { | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3931 assert(_cm->out_of_regions(), |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3932 "at this point we should be out of regions"); |
342 | 3933 } |
3934 } while ( _curr_region != NULL && !has_aborted()); | |
3935 | |
3936 if (!has_aborted()) { | |
3937 // We cannot check whether the global stack is empty, since other | |
343
afc1ce1efe66
6710665: G1: guarantee(_cm->out_of_regions() && _cm->region_stack_empty() && _task_queue->size() == 0, ...)
iveresov
parents:
342
diff
changeset
|
3938 // tasks might be pushing objects to it concurrently. We also cannot |
afc1ce1efe66
6710665: G1: guarantee(_cm->out_of_regions() && _cm->region_stack_empty() && _task_queue->size() == 0, ...)
iveresov
parents:
342
diff
changeset
|
3939 // check if the region stack is empty because if a thread is aborting |
afc1ce1efe66
6710665: G1: guarantee(_cm->out_of_regions() && _cm->region_stack_empty() && _task_queue->size() == 0, ...)
iveresov
parents:
342
diff
changeset
|
3940 // it can push a partially done region back. |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3941 assert(_cm->out_of_regions(), |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3942 "at this point we should be out of regions"); |
342 | 3943 |
3944 if (_cm->verbose_low()) | |
3945 gclog_or_tty->print_cr("[%d] all regions claimed", _task_id); | |
3946 | |
3947 // Try to reduce the number of available SATB buffers so that | |
3948 // remark has less work to do. | |
3949 drain_satb_buffers(); | |
3950 } | |
3951 | |
3952 // Since we've done everything else, we can now totally drain the | |
3953 // local queue and global stack. | |
3954 drain_local_queue(false); | |
3955 drain_global_stack(false); | |
3956 | |
3957 // Attempt at work stealing from other task's queues. | |
3958 if (!has_aborted()) { | |
3959 // We have not aborted. This means that we have finished all that | |
3960 // we could. Let's try to do some stealing... | |
3961 | |
3962 // We cannot check whether the global stack is empty, since other | |
343
afc1ce1efe66
6710665: G1: guarantee(_cm->out_of_regions() && _cm->region_stack_empty() && _task_queue->size() == 0, ...)
iveresov
parents:
342
diff
changeset
|
3963 // tasks might be pushing objects to it concurrently. We also cannot |
afc1ce1efe66
6710665: G1: guarantee(_cm->out_of_regions() && _cm->region_stack_empty() && _task_queue->size() == 0, ...)
iveresov
parents:
342
diff
changeset
|
3964 // check if the region stack is empty because if a thread is aborting |
afc1ce1efe66
6710665: G1: guarantee(_cm->out_of_regions() && _cm->region_stack_empty() && _task_queue->size() == 0, ...)
iveresov
parents:
342
diff
changeset
|
3965 // it can push a partially done region back. |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3966 assert(_cm->out_of_regions() && _task_queue->size() == 0, |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3967 "only way to reach here"); |
342 | 3968 |
3969 if (_cm->verbose_low()) | |
3970 gclog_or_tty->print_cr("[%d] starting to steal", _task_id); | |
3971 | |
3972 while (!has_aborted()) { | |
3973 oop obj; | |
3974 statsOnly( ++_steal_attempts ); | |
3975 | |
3976 if (_cm->try_stealing(_task_id, &_hash_seed, obj)) { | |
3977 if (_cm->verbose_medium()) | |
3978 gclog_or_tty->print_cr("[%d] stolen "PTR_FORMAT" successfully", | |
3979 _task_id, (void*) obj); | |
3980 | |
3981 statsOnly( ++_steals ); | |
3982 | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3983 assert(_nextMarkBitMap->isMarked((HeapWord*) obj), |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
3984 "any stolen object should be marked"); |
342 | 3985 scan_object(obj); |
3986 | |
3987 // And since we're towards the end, let's totally drain the | |
3988 // local queue and global stack. | |
3989 drain_local_queue(false); | |
3990 drain_global_stack(false); | |
3991 } else { | |
3992 break; | |
3993 } | |
3994 } | |
3995 } | |
3996 | |
3997 // We still haven't aborted. Now, let's try to get into the | |
3998 // termination protocol. | |
3999 if (!has_aborted()) { | |
4000 // We cannot check whether the global stack is empty, since other | |
343
afc1ce1efe66
6710665: G1: guarantee(_cm->out_of_regions() && _cm->region_stack_empty() && _task_queue->size() == 0, ...)
iveresov
parents:
342
diff
changeset
|
4001 // tasks might be concurrently pushing objects on it. We also cannot |
afc1ce1efe66
6710665: G1: guarantee(_cm->out_of_regions() && _cm->region_stack_empty() && _task_queue->size() == 0, ...)
iveresov
parents:
342
diff
changeset
|
4002 // check if the region stack is empty because if a thread is aborting |
afc1ce1efe66
6710665: G1: guarantee(_cm->out_of_regions() && _cm->region_stack_empty() && _task_queue->size() == 0, ...)
iveresov
parents:
342
diff
changeset
|
4003 // it can push a partially done region back. |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
4004 // Separated the asserts so that we know which one fires. |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
4005 assert(_cm->out_of_regions(), "only way to reach here"); |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
4006 assert(_task_queue->size() == 0, "only way to reach here"); |
342 | 4007 |
4008 if (_cm->verbose_low()) | |
4009 gclog_or_tty->print_cr("[%d] starting termination protocol", _task_id); | |
4010 | |
4011 _termination_start_time_ms = os::elapsedVTime() * 1000.0; | |
4012 // The CMTask class also extends the TerminatorTerminator class, | |
4013 // hence its should_exit_termination() method will also decide | |
4014 // whether to exit the termination protocol or not. | |
4015 bool finished = _cm->terminator()->offer_termination(this); | |
4016 double termination_end_time_ms = os::elapsedVTime() * 1000.0; | |
4017 _termination_time_ms += | |
4018 termination_end_time_ms - _termination_start_time_ms; | |
4019 | |
4020 if (finished) { | |
4021 // We're all done. | |
4022 | |
4023 if (_task_id == 0) { | |
4024 // let's allow task 0 to do this | |
4025 if (concurrent()) { | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
4026 assert(_cm->concurrent_marking_in_progress(), "invariant"); |
342 | 4027 // we need to set this to false before the next |
4028 // safepoint. This way we ensure that the marking phase | |
4029 // doesn't observe any more heap expansions. | |
4030 _cm->clear_concurrent_marking_in_progress(); | |
4031 } | |
4032 } | |
4033 | |
4034 // We can now guarantee that the global stack is empty, since | |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
4035 // all other tasks have finished. We separated the guarantees so |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
4036 // that, if a condition is false, we can immediately find out |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
4037 // which one. |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
4038 guarantee(_cm->out_of_regions(), "only way to reach here"); |
1835
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
4039 guarantee(_aborted_region.is_empty(), "only way to reach here"); |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
4040 guarantee(_cm->region_stack_empty(), "only way to reach here"); |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
4041 guarantee(_cm->mark_stack_empty(), "only way to reach here"); |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
4042 guarantee(_task_queue->size() == 0, "only way to reach here"); |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
4043 guarantee(!_cm->has_overflown(), "only way to reach here"); |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
4044 guarantee(!_cm->mark_stack_overflow(), "only way to reach here"); |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
4045 guarantee(!_cm->region_stack_overflow(), "only way to reach here"); |
342 | 4046 |
4047 if (_cm->verbose_low()) | |
4048 gclog_or_tty->print_cr("[%d] all tasks terminated", _task_id); | |
4049 } else { | |
4050 // Apparently there's more work to do. Let's abort this task. It | |
4051 // will restart it and we can hopefully find more things to do. | |
4052 | |
4053 if (_cm->verbose_low()) | |
4054 gclog_or_tty->print_cr("[%d] apparently there is more work to do", _task_id); | |
4055 | |
4056 set_has_aborted(); | |
4057 statsOnly( ++_aborted_termination ); | |
4058 } | |
4059 } | |
4060 | |
4061 // Mainly for debugging purposes to make sure that a pointer to the | |
4062 // closure which was statically allocated in this frame doesn't | |
4063 // escape it by accident. | |
4064 set_oop_closure(NULL); | |
4065 double end_time_ms = os::elapsedVTime() * 1000.0; | |
4066 double elapsed_time_ms = end_time_ms - _start_time_ms; | |
4067 // Update the step history. | |
4068 _step_times_ms.add(elapsed_time_ms); | |
4069 | |
4070 if (has_aborted()) { | |
4071 // The task was aborted for some reason. | |
4072 | |
4073 statsOnly( ++_aborted ); | |
4074 | |
4075 if (_has_aborted_timed_out) { | |
4076 double diff_ms = elapsed_time_ms - _time_target_ms; | |
4077 // Keep statistics of how well we did with respect to hitting | |
4078 // our target only if we actually timed out (if we aborted for | |
4079 // other reasons, then the results might get skewed). | |
4080 _marking_step_diffs_ms.add(diff_ms); | |
4081 } | |
4082 | |
4083 if (_cm->has_overflown()) { | |
4084 // This is the interesting one. We aborted because a global | |
4085 // overflow was raised. This means we have to restart the | |
4086 // marking phase and start iterating over regions. However, in | |
4087 // order to do this we have to make sure that all tasks stop | |
4088 // what they are doing and re-initialise in a safe manner. We | |
4089 // will achieve this with the use of two barrier sync points. | |
4090 | |
4091 if (_cm->verbose_low()) | |
4092 gclog_or_tty->print_cr("[%d] detected overflow", _task_id); | |
4093 | |
4094 _cm->enter_first_sync_barrier(_task_id); | |
4095 // When we exit this sync barrier we know that all tasks have | |
4096 // stopped doing marking work. So, it's now safe to | |
4097 // re-initialise our data structures. At the end of this method, | |
4098 // task 0 will clear the global data structures. | |
4099 | |
4100 statsOnly( ++_aborted_overflow ); | |
4101 | |
4102 // We clear the local state of this task... | |
4103 clear_region_fields(); | |
4104 | |
4105 // ...and enter the second barrier. | |
4106 _cm->enter_second_sync_barrier(_task_id); | |
4107 // At this point everything has bee re-initialised and we're | |
4108 // ready to restart. | |
4109 } | |
4110 | |
4111 if (_cm->verbose_low()) { | |
4112 gclog_or_tty->print_cr("[%d] <<<<<<<<<< ABORTING, target = %1.2lfms, " | |
4113 "elapsed = %1.2lfms <<<<<<<<<<", | |
4114 _task_id, _time_target_ms, elapsed_time_ms); | |
4115 if (_cm->has_aborted()) | |
4116 gclog_or_tty->print_cr("[%d] ========== MARKING ABORTED ==========", | |
4117 _task_id); | |
4118 } | |
4119 } else { | |
4120 if (_cm->verbose_low()) | |
4121 gclog_or_tty->print_cr("[%d] <<<<<<<<<< FINISHED, target = %1.2lfms, " | |
4122 "elapsed = %1.2lfms <<<<<<<<<<", | |
4123 _task_id, _time_target_ms, elapsed_time_ms); | |
4124 } | |
4125 | |
4126 _claimed = false; | |
4127 } | |
4128 | |
4129 CMTask::CMTask(int task_id, | |
4130 ConcurrentMark* cm, | |
4131 CMTaskQueue* task_queue, | |
4132 CMTaskQueueSet* task_queues) | |
4133 : _g1h(G1CollectedHeap::heap()), | |
4134 _task_id(task_id), _cm(cm), | |
4135 _claimed(false), | |
4136 _nextMarkBitMap(NULL), _hash_seed(17), | |
4137 _task_queue(task_queue), | |
4138 _task_queues(task_queues), | |
1835
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
4139 _oop_closure(NULL), |
4805b9f4779e
6941395: G1: Use only lock-free versions of region stack push() and pop()
johnc
parents:
1833
diff
changeset
|
4140 _aborted_region(MemRegion()) { |
1023
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
4141 guarantee(task_queue != NULL, "invariant"); |
11d4857fe5e1
6888619: G1: too many guarantees in concurrent marking
tonyp
parents:
1022
diff
changeset
|
4142 guarantee(task_queues != NULL, "invariant"); |
342 | 4143 |
4144 statsOnly( _clock_due_to_scanning = 0; | |
4145 _clock_due_to_marking = 0 ); | |
4146 | |
4147 _marking_step_diffs_ms.add(0.5); | |
4148 } |