/* * Copyright (c) 2015, 2017, Red Hat, Inc. and/or its affiliates. * * This code is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License version 2 only, as * published by the Free Software Foundation. * * This code is distributed in the hope that it will be useful, but WITHOUT * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * version 2 for more details (a copy is included in the LICENSE file that * accompanied this code). * * You should have received a copy of the GNU General Public License version * 2 along with this work; if not, write to the Free Software Foundation, * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA. * * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA * or visit www.oracle.com if you need additional information or have any * questions. * */ #ifndef SHARE_VM_GC_SHENANDOAH_SHENANDOAHHEAP_INLINE_HPP #define SHARE_VM_GC_SHENANDOAH_SHENANDOAHHEAP_INLINE_HPP #include "classfile/javaClasses.inline.hpp" #include "gc/g1/suspendibleThreadSet.hpp" #include "gc/shared/markBitMap.inline.hpp" #include "gc/shared/threadLocalAllocBuffer.inline.hpp" #include "gc/shenandoah/brooksPointer.inline.hpp" #include "gc/shenandoah/shenandoahBarrierSet.inline.hpp" #include "gc/shenandoah/shenandoahCollectionSet.hpp" #include "gc/shenandoah/shenandoahCollectionSet.inline.hpp" #include "gc/shenandoah/shenandoahConnectionMatrix.inline.hpp" #include "gc/shenandoah/shenandoahHeap.hpp" #include "gc/shenandoah/shenandoahHeapRegionSet.hpp" #include "gc/shenandoah/shenandoahHeapRegion.inline.hpp" #include "gc/shenandoah/shenandoahStringDedup.hpp" #include "gc/shenandoah/shenandoahUtils.hpp" #include "oops/oop.inline.hpp" #include "runtime/atomic.hpp" #include "runtime/interfaceSupport.hpp" #include "runtime/prefetch.hpp" #include "runtime/prefetch.inline.hpp" #include "runtime/thread.hpp" #include "utilities/copy.hpp" template void ShenandoahUpdateRefsClosure::do_oop_work(T* p) { T o = oopDesc::load_heap_oop(p); if (! oopDesc::is_null(o)) { oop obj = oopDesc::decode_heap_oop_not_null(o); _heap->update_oop_ref_not_null(p, obj); } } void ShenandoahUpdateRefsClosure::do_oop(oop* p) { do_oop_work(p); } void ShenandoahUpdateRefsClosure::do_oop(narrowOop* p) { do_oop_work(p); } /* * Marks the object. Returns true if the object has not been marked before and has * been marked by this thread. Returns false if the object has already been marked, * or if a competing thread succeeded in marking this object. */ inline bool ShenandoahHeap::mark(oop obj) { #ifdef ASSERT if (! oopDesc::unsafe_equals(obj, oopDesc::bs()->read_barrier(obj))) { tty->print_cr("heap region containing obj:"); ShenandoahHeapRegion* obj_region = heap_region_containing(obj); obj_region->print(); tty->print_cr("heap region containing forwardee:"); ShenandoahHeapRegion* forward_region = heap_region_containing(oopDesc::bs()->read_barrier(obj)); forward_region->print(); } #endif assert(oopDesc::unsafe_equals(obj, oopDesc::bs()->read_barrier(obj)), "only mark forwarded copy of objects"); return mark_no_checks(obj); } inline bool ShenandoahHeap::mark_no_checks(oop obj) { HeapWord* addr = (HeapWord*) obj; return !allocated_after_mark_start(addr) && _mark_bit_map.parMark(addr); } inline bool ShenandoahHeap::is_marked(oop obj) const { HeapWord* addr = (HeapWord*) obj; return allocated_after_mark_start(addr) || _mark_bit_map.isMarked(addr); } inline bool ShenandoahHeap::need_update_refs() const { return _need_update_refs; } inline size_t ShenandoahHeap::heap_region_index_containing(const void* addr) const { uintptr_t region_start = ((uintptr_t) addr); uintptr_t index = (region_start - (uintptr_t) base()) >> ShenandoahHeapRegion::region_size_bytes_shift(); #ifdef ASSERT if (index >= num_regions()) { tty->print_cr("heap region does not contain address, heap base: "PTR_FORMAT \ ", real bottom of first region: "PTR_FORMAT", num_regions: "SIZE_FORMAT", region_size: "SIZE_FORMAT, p2i(base()), p2i(_ordered_regions->get(0)->bottom()), num_regions(), ShenandoahHeapRegion::region_size_bytes()); } #endif assert(index < num_regions(), "heap region index must be in range"); return index; } inline ShenandoahHeapRegion* ShenandoahHeap::heap_region_containing(const void* addr) const { size_t index = heap_region_index_containing(addr); ShenandoahHeapRegion* result = _ordered_regions->get(index); #ifdef ASSERT if (!(addr >= result->bottom() && addr < result->end())) { tty->print_cr("heap region does not contain address, heap base: "PTR_FORMAT \ ", real bottom of first region: "PTR_FORMAT", num_regions: "SIZE_FORMAT, p2i(base()), p2i(_ordered_regions->get(0)->bottom()), num_regions()); } #endif assert(addr >= result->bottom() && addr < result->end(), "address must be in found region"); return result; } template inline oop ShenandoahHeap::update_oop_ref_not_null(T* p, oop obj) { if (in_collection_set(obj)) { oop forw = ShenandoahBarrierSet::resolve_oop_static_not_null(obj); assert(! oopDesc::unsafe_equals(forw, obj) || is_full_gc_in_progress() || cancelled_concgc(), "expect forwarded object"); obj = forw; oopDesc::encode_store_heap_oop(p, obj); } #ifdef ASSERT else { assert(oopDesc::unsafe_equals(obj, ShenandoahBarrierSet::resolve_oop_static_not_null(obj)), "expect not forwarded"); } #endif return obj; } template inline oop ShenandoahHeap::maybe_update_oop_ref(T* p) { T o = oopDesc::load_heap_oop(p); if (! oopDesc::is_null(o)) { oop obj = oopDesc::decode_heap_oop_not_null(o); return maybe_update_oop_ref_not_null(p, obj); } else { return NULL; } } template inline oop ShenandoahHeap::evac_update_oop_ref(T* p, bool& evac) { evac = false; T o = oopDesc::load_heap_oop(p); if (! oopDesc::is_null(o)) { oop heap_oop = oopDesc::decode_heap_oop_not_null(o); if (in_collection_set(heap_oop)) { oop forwarded_oop = ShenandoahBarrierSet::resolve_oop_static_not_null(heap_oop); // read brooks ptr if (oopDesc::unsafe_equals(forwarded_oop, heap_oop)) { forwarded_oop = evacuate_object(heap_oop, Thread::current(), evac); } oop prev = atomic_compare_exchange_oop(forwarded_oop, p, heap_oop); if (prev == heap_oop) { return forwarded_oop; } else { return NULL; } } return heap_oop; } else { return NULL; } } inline oop ShenandoahHeap::atomic_compare_exchange_oop(oop n, oop* addr, oop c) { return (oop) Atomic::cmpxchg_ptr(n, addr, c); } inline oop ShenandoahHeap::atomic_compare_exchange_oop(oop n, narrowOop* addr, oop c) { narrowOop cmp = oopDesc::encode_heap_oop(c); narrowOop val = oopDesc::encode_heap_oop(n); return oopDesc::decode_heap_oop((narrowOop) Atomic::cmpxchg(val, addr, cmp)); } template inline oop ShenandoahHeap::maybe_update_oop_ref_not_null(T* p, oop heap_oop) { assert((! is_in(p)) || (! in_collection_set(p)) || is_full_gc_in_progress(), "never update refs in from-space, unless evacuation has been cancelled"); #ifdef ASSERT if (! is_in(heap_oop)) { print_heap_regions_on(tty); tty->print_cr("object not in heap: "PTR_FORMAT", referenced by: "PTR_FORMAT, p2i((HeapWord*) heap_oop), p2i(p)); assert(is_in(heap_oop), "object must be in heap"); } #endif assert(is_in(heap_oop), "only ever call this on objects in the heap"); if (in_collection_set(heap_oop)) { oop forwarded_oop = ShenandoahBarrierSet::resolve_oop_static_not_null(heap_oop); // read brooks ptr if (oopDesc::unsafe_equals(forwarded_oop, heap_oop)) { // E.g. during evacuation. return forwarded_oop; } assert(! oopDesc::unsafe_equals(forwarded_oop, heap_oop) || is_full_gc_in_progress(), "expect forwarded object"); log_develop_trace(gc)("Updating old ref: "PTR_FORMAT" pointing to "PTR_FORMAT" to new ref: "PTR_FORMAT, p2i(p), p2i(heap_oop), p2i(forwarded_oop)); assert(oopDesc::is_oop(forwarded_oop), "oop required"); assert(is_in(forwarded_oop), "forwardee must be in heap"); assert(oopDesc::bs()->is_safe(forwarded_oop), "forwardee must not be in collection set"); // If this fails, another thread wrote to p before us, it will be logged in SATB and the // reference be updated later. oop result = atomic_compare_exchange_oop(forwarded_oop, p, heap_oop); if (oopDesc::unsafe_equals(result, heap_oop)) { // CAS successful. return forwarded_oop; } else { // Note: we used to assert the following here. This doesn't work because sometimes, during // marking/updating-refs, it can happen that a Java thread beats us with an arraycopy, // which first copies the array, which potentially contains from-space refs, and only afterwards // updates all from-space refs to to-space refs, which leaves a short window where the new array // elements can be from-space. // assert(oopDesc::is_null(result) || // oopDesc::unsafe_equals(result, ShenandoahBarrierSet::resolve_oop_static_not_null(result)), // "expect not forwarded"); return NULL; } } else { assert(oopDesc::unsafe_equals(heap_oop, ShenandoahBarrierSet::resolve_oop_static_not_null(heap_oop)), "expect not forwarded"); return heap_oop; } } inline bool ShenandoahHeap::cancelled_concgc() const { return OrderAccess::load_acquire((jbyte*) &_cancelled_concgc) == CANCELLED; } inline bool ShenandoahHeap::check_cancelled_concgc_and_yield(bool sts_active) { if (! (sts_active && ShenandoahSuspendibleWorkers)) { return cancelled_concgc(); } jbyte prev = Atomic::cmpxchg((jbyte)NOT_CANCELLED, &_cancelled_concgc, (jbyte)CANCELLABLE); if (prev == CANCELLABLE || prev == NOT_CANCELLED) { if (SuspendibleThreadSet::should_yield()) { SuspendibleThreadSet::yield(); } // Back to CANCELLABLE. The thread that poked NOT_CANCELLED first gets // to restore to CANCELLABLE. if (prev == CANCELLABLE) { OrderAccess::release_store_fence(&_cancelled_concgc, CANCELLABLE); } return false; } else { return true; } } inline bool ShenandoahHeap::try_cancel_concgc() { while (true) { jbyte prev = Atomic::cmpxchg((jbyte)CANCELLED, &_cancelled_concgc, (jbyte)CANCELLABLE); if (prev == CANCELLABLE) return true; else if (prev == CANCELLED) return false; assert(ShenandoahSuspendibleWorkers, "should not get here when not using suspendible workers"); assert(prev == NOT_CANCELLED, "must be NOT_CANCELLED"); { // We need to provide a safepoint here, otherwise we might // spin forever if a SP is pending. ThreadBlockInVM sp(JavaThread::current()); SpinPause(); } } } inline void ShenandoahHeap::clear_cancelled_concgc() { OrderAccess::release_store_fence(&_cancelled_concgc, CANCELLABLE); } inline HeapWord* ShenandoahHeap::allocate_from_gclab(Thread* thread, size_t size) { if (UseTLAB) { if (!thread->gclab().is_initialized()) { assert(!thread->is_Java_thread() && !thread->is_Worker_thread(), "Performance: thread should have GCLAB: %s", thread->name()); // No GCLABs in this thread, fallback to shared allocation return NULL; } HeapWord* obj = thread->gclab().allocate(size); if (obj != NULL) { return obj; } // Otherwise... return allocate_from_gclab_slow(thread, size); } else { return NULL; } } inline oop ShenandoahHeap::evacuate_object(oop p, Thread* thread, bool& evacuated) { evacuated = false; size_t size_no_fwdptr = (size_t) p->size(); size_t size_with_fwdptr = size_no_fwdptr + BrooksPointer::word_size(); assert(!heap_region_containing(p)->is_humongous(), "never evacuate humongous objects"); bool alloc_from_gclab = true; HeapWord* filler = allocate_from_gclab(thread, size_with_fwdptr); if (filler == NULL) { filler = allocate_memory(size_with_fwdptr, _alloc_shared_gc); alloc_from_gclab = false; } #ifdef ASSERT // Checking that current Java thread does not hold Threads_lock when we get here. // If that ever be the case, we'd deadlock in oom_during_evacuation. if ((! Thread::current()->is_GC_task_thread()) && (! Thread::current()->is_ConcurrentGC_thread())) { assert(! Threads_lock->owned_by_self() || SafepointSynchronize::is_at_safepoint(), "must not hold Threads_lock here"); } #endif if (filler == NULL) { oom_during_evacuation(); // If this is a Java thread, it should have waited // until all GC threads are done, and then we // return the forwardee. oop resolved = ShenandoahBarrierSet::resolve_oop_static(p); return resolved; } // Copy the object and initialize its forwarding ptr: HeapWord* copy = filler + BrooksPointer::word_size(); oop copy_val = oop(copy); Copy::aligned_disjoint_words((HeapWord*) p, copy, size_no_fwdptr); BrooksPointer::initialize(oop(copy)); log_develop_trace(gc, compaction)("Copy object: " PTR_FORMAT " -> " PTR_FORMAT, p2i(p), p2i(copy)); // String dedup support bool need_str_dedup = false; if (ShenandoahStringDedup::is_enabled() && java_lang_String::is_instance_inlined(copy_val)) { // We need to increase age before CAS to avoid race condition. // Once new copy is published, other threads may set hash code, // or perform locking, etc. which will race age bits manipulation. copy_val->incr_age(); need_str_dedup = ShenandoahStringDedup::is_candidate(copy_val); } // Try to install the new forwarding pointer. oop result = BrooksPointer::try_update_forwardee(p, copy_val); if (oopDesc::unsafe_equals(result, p)) { // Successfully evacuated. Our copy is now the public one! evacuated = true; log_develop_trace(gc, compaction)("Copy object: " PTR_FORMAT " -> " PTR_FORMAT " succeeded", p2i(p), p2i(copy)); // Only dedup evacuated string if (need_str_dedup) { // Shenandoah evacuates objects inside and outside of safepoints. // But string dedup protocol requires deduplication outside of safepoints, // so we need to queue candidates during safepoints. if (SafepointSynchronize::is_at_safepoint()) { assert(thread->is_Worker_thread(), "Must be a worker thread during a safepoint"); // Use worker thread id instead of worker_id to avoid passing down worker_id. // This may cause imbalance among the queues, but it is okay, since deduplication is // single threaded. ShenandoahStringDedup::enqueue_from_safepoint(copy_val, thread->as_Worker_thread()->id()); } else { ShenandoahStringDedup::deduplicate(copy_val); } } #ifdef ASSERT assert(oopDesc::is_oop(copy_val), "expect oop"); assert(p->klass() == copy_val->klass(), "Should have the same class p: "PTR_FORMAT", copy: "PTR_FORMAT, p2i(p), p2i(copy)); #endif return copy_val; } else { // Failed to evacuate. We need to deal with the object that is left behind. Since this // new allocation is certainly after TAMS, it will be considered live in the next cycle. // But if it happens to contain references to evacuated regions, those references would // not get updated for this stale copy during this cycle, and we will crash while scanning // it the next cycle. // // For GCLAB allocations, it is enough to rollback the allocation ptr. Either the next // object will overwrite this stale copy, or the filler object on LAB retirement will // do this. For non-GCLAB allocations, we have no way to retract the allocation, and // have to explicitly overwrite the copy with the filler object. With that overwrite, // we have to keep the fwdptr initialized and pointing to our (stale) copy. if (alloc_from_gclab) { thread->gclab().rollback(size_with_fwdptr); } else { fill_with_object(copy, size_no_fwdptr); } log_develop_trace(gc, compaction)("Copy object: " PTR_FORMAT " -> " PTR_FORMAT " failed, use other: " PTR_FORMAT, p2i(p), p2i(copy), p2i(result)); return result; } } inline bool ShenandoahHeap::requires_marking(const void* entry) const { // TODO: Make this faster! It's used in a hot path. // TODO: it's not strictly matrix-related, but used only in partial (i.e. matrix) GCs. if (is_concurrent_partial_in_progress()) { assert(! in_collection_set((oop) entry), "must not get cset objects here"); // assert(free_regions()->contains(heap_region_containing(entry)), "expect to-space object"); return true; } else if (concurrent_mark_in_progress()) { return ! is_marked(oop(entry)); } else { return false; } } bool ShenandoahHeap::region_in_collection_set(size_t region_index) const { assert(collection_set() != NULL, "Sanity"); return collection_set()->is_in(region_index); } bool ShenandoahHeap::in_collection_set(ShenandoahHeapRegion* r) const { return region_in_collection_set(r->region_number()); } template inline bool ShenandoahHeap::in_collection_set(T p) const { HeapWord* obj = (HeapWord*) p; assert(collection_set() != NULL, "Sanity"); assert(is_in(obj), "should be in heap"); return collection_set()->is_in(obj); } inline bool ShenandoahHeap::concurrent_mark_in_progress() const { return _concurrent_mark_in_progress != 0; } inline bool ShenandoahHeap::is_concurrent_partial_in_progress() const { return _concurrent_partial_in_progress; } inline address ShenandoahHeap::update_refs_in_progress_addr() { return (address) &(ShenandoahHeap::heap()->_update_refs_in_progress); } inline bool ShenandoahHeap::is_evacuation_in_progress() const { return _evacuation_in_progress != 0; } inline address ShenandoahHeap::evacuation_in_progress_addr() { return (address) &(ShenandoahHeap::heap()->_evacuation_in_progress); } inline bool ShenandoahHeap::allocated_after_mark_start(HeapWord* addr) const { uintx index = ((uintx) addr) >> ShenandoahHeapRegion::region_size_bytes_shift(); HeapWord* top_at_mark_start = _top_at_mark_starts[index]; bool alloc_after_mark_start = addr >= top_at_mark_start; return alloc_after_mark_start; } template inline void ShenandoahHeap::marked_object_iterate(ShenandoahHeapRegion* region, T* cl) { marked_object_iterate(region, cl, region->top()); } template inline void ShenandoahHeap::marked_object_safe_iterate(ShenandoahHeapRegion* region, T* cl) { marked_object_iterate(region, cl, region->concurrent_iteration_safe_limit()); } template inline void ShenandoahHeap::marked_object_iterate(ShenandoahHeapRegion* region, T* cl, HeapWord* limit) { assert(BrooksPointer::word_offset() < 0, "skip_delta calculation below assumes the forwarding ptr is before obj"); assert(! region->is_humongous_continuation(), "no humongous continuation regions here"); assert(is_bitmap_valid(), "only try this with complete marking bitmap"); MarkBitMap mark_bit_map = _mark_bit_map; HeapWord* tams = top_at_mark_start(region->bottom()); size_t skip_bitmap_delta = BrooksPointer::word_size() + 1; size_t skip_objsize_delta = BrooksPointer::word_size() /* + actual obj.size() below */; HeapWord* start = region->bottom() + BrooksPointer::word_size(); HeapWord* end = MIN2(tams + BrooksPointer::word_size(), region->end()); HeapWord* addr = mark_bit_map.getNextMarkedWordAddress(start, end); intx dist = ShenandoahMarkScanPrefetch; if (dist > 0) { // Batched scan that prefetches the oop data, anticipating the access to // either header, oop field, or forwarding pointer. Not that we cannot // touch anything in oop, while it still being prefetched to get enough // time for prefetch to work. This is why we try to scan the bitmap linearly, // disregarding the object size. However, since we know forwarding pointer // preceeds the object, we can skip over it. Once we cannot trust the bitmap, // there is no point for prefetching the oop contents, as oop->size() will // touch it prematurely. // No variable-length arrays in standard C++, have enough slots to fit // the prefetch distance. static const int SLOT_COUNT = 256; guarantee(dist <= SLOT_COUNT, "adjust slot count"); oop slots[SLOT_COUNT]; bool aborting = false; int avail; do { avail = 0; for (int c = 0; (c < dist) && (addr < limit); c++) { Prefetch::read(addr, BrooksPointer::byte_offset()); oop obj = oop(addr); slots[avail++] = obj; if (addr < tams) { addr += skip_bitmap_delta; addr = mark_bit_map.getNextMarkedWordAddress(addr, end); } else { // cannot trust mark bitmap anymore, finish the current stride, // and switch to accurate traversal addr += obj->size() + skip_objsize_delta; aborting = true; } } for (int c = 0; c < avail; c++) { do_marked_object(cl, slots[c]); } } while (avail > 0 && !aborting); // accurate traversal while (addr < limit) { oop obj = oop(addr); int size = obj->size(); do_marked_object(cl, obj); addr += size + skip_objsize_delta; } } else { while (addr < limit) { oop obj = oop(addr); int size = obj->size(); do_marked_object(cl, obj); addr += size + skip_objsize_delta; if (addr < tams) { addr = mark_bit_map.getNextMarkedWordAddress(addr, end); } } } } template inline void ShenandoahHeap::do_marked_object(T* cl, oop obj) { assert(!oopDesc::is_null(obj), "sanity"); assert(oopDesc::is_oop(obj), "sanity"); assert(is_in(obj), "sanity"); assert(is_marked(obj), "object expected to be marked"); cl->do_object(obj); } template class ShenandoahObjectToOopClosure : public ObjectClosure { T* _cl; public: ShenandoahObjectToOopClosure(T* cl) : _cl(cl) {} void do_object(oop obj) { obj->oop_iterate(_cl); } }; template class ShenandoahObjectToOopBoundedClosure : public ObjectClosure { T* _cl; MemRegion _bounds; public: ShenandoahObjectToOopBoundedClosure(T* cl, HeapWord* bottom, HeapWord* top) : _cl(cl), _bounds(bottom, top) {} void do_object(oop obj) { obj->oop_iterate(_cl, _bounds); } }; template inline void ShenandoahHeap::marked_object_oop_iterate(ShenandoahHeapRegion* region, T* cl, HeapWord* top) { if (region->is_humongous()) { HeapWord* bottom = region->bottom(); if (top > bottom) { region = region->humongous_start_region(); ShenandoahObjectToOopBoundedClosure objs(cl, bottom, top); marked_object_iterate(region, &objs); } } else { ShenandoahObjectToOopClosure objs(cl); marked_object_iterate(region, &objs, top); } } template inline void ShenandoahHeap::marked_object_oop_iterate(ShenandoahHeapRegion* region, T* cl) { marked_object_oop_iterate(region, cl, region->top()); } template inline void ShenandoahHeap::marked_object_oop_safe_iterate(ShenandoahHeapRegion* region, T* cl) { marked_object_oop_iterate(region, cl, region->concurrent_iteration_safe_limit()); } #endif // SHARE_VM_GC_SHENANDOAH_SHENANDOAHHEAP_INLINE_HPP