graal-jvmci-8: src/share/vm/runtime/sweeper.cpp comparison

comparison src/share/vm/runtime/sweeper.cpp @ 12324:510fbd28919c

8020151: PSR:PERF Large performance regressions when code cache is filled Summary: Code cache sweeping based on method hotness; removed speculatively disconnect Reviewed-by: kvn, iveresov

author	anoll
date	Fri, 27 Sep 2013 10:50:55 +0200
parents	ab274453d37f
children	cefad50507d8 78da3894b86f

comparison

equal deleted inserted replaced

-:c9ccd7b85f20
+:510fbd28919c
 }
 #else
 #define SWEEP(nm)
 #endif
+nmethod*  NMethodSweeper::_current         = NULL; // Current nmethod
-long      NMethodSweeper::_traversals = 0;   // No. of stack traversals performed
+long      NMethodSweeper::_traversals      = 0;    // Nof. stack traversals performed
-nmethod*  NMethodSweeper::_current = NULL;   // Current nmethod
+int       NMethodSweeper::_seen            = 0;    // Nof. nmethods we have currently processed in current pass of CodeCache
-int       NMethodSweeper::_seen = 0 ;        // No. of nmethods we have currently processed in current pass of CodeCache
+int       NMethodSweeper::_flushed_count   = 0;    // Nof. nmethods flushed in current sweep
-int       NMethodSweeper::_flushed_count = 0;   // Nof. nmethods flushed in current sweep
+int       NMethodSweeper::_zombified_count = 0;    // Nof. nmethods made zombie in current sweep
-int       NMethodSweeper::_zombified_count = 0; // Nof. nmethods made zombie in current sweep
+int       NMethodSweeper::_marked_count    = 0;    // Nof. nmethods marked for reclaim in current sweep
-int       NMethodSweeper::_marked_count = 0;    // Nof. nmethods marked for reclaim in current sweep
+volatile int NMethodSweeper::_invocations   = 0; // Nof. invocations left until we are completed with this pass
-volatile int NMethodSweeper::_invocations = 0;   // No. of invocations left until we are completed with this pass
 volatile int NMethodSweeper::_sweep_started = 0; // Whether a sweep is in progress.
-jint      NMethodSweeper::_locked_seen = 0;
+jint      NMethodSweeper::_locked_seen               = 0;
 jint      NMethodSweeper::_not_entrant_seen_on_stack = 0;
-bool      NMethodSweeper::_resweep = false;
+bool      NMethodSweeper::_request_mark_phase        = false;
-jint      NMethodSweeper::_flush_token = 0;
-jlong     NMethodSweeper::_last_full_flush_time = 0;
-int       NMethodSweeper::_highest_marked = 0;
-int       NMethodSweeper::_dead_compile_ids = 0;
-long      NMethodSweeper::_last_flush_traversal_id = 0;
-int       NMethodSweeper::_number_of_flushes = 0; // Total of full traversals caused by full cache
 int       NMethodSweeper::_total_nof_methods_reclaimed = 0;
-jlong     NMethodSweeper::_total_time_sweeping = 0;
+jlong     NMethodSweeper::_total_time_sweeping         = 0;
-jlong     NMethodSweeper::_total_time_this_sweep = 0;
+jlong     NMethodSweeper::_total_time_this_sweep       = 0;
-jlong     NMethodSweeper::_peak_sweep_time = 0;
+jlong     NMethodSweeper::_peak_sweep_time             = 0;
-jlong     NMethodSweeper::_peak_sweep_fraction_time = 0;
+jlong     NMethodSweeper::_peak_sweep_fraction_time    = 0;
-jlong     NMethodSweeper::_total_disconnect_time = 0;
+int       NMethodSweeper::_hotness_counter_reset_val   = 0;
-jlong     NMethodSweeper::_peak_disconnect_time = 0;
 class MarkActivationClosure: public CodeBlobClosure {
 public:
 virtual void do_code_blob(CodeBlob* cb) {
-// If we see an activation belonging to a non_entrant nmethod, we mark it.
+if (cb->is_nmethod()) {
-if (cb->is_nmethod() && ((nmethod*)cb)->is_not_entrant()) {
+nmethod* nm = (nmethod*)cb;
-((nmethod*)cb)->mark_as_seen_on_stack();
+nm->set_hotness_counter(NMethodSweeper::hotness_counter_reset_val());
+// If we see an activation belonging to a non_entrant nmethod, we mark it.
+if (nm->is_not_entrant()) {
+nm->mark_as_seen_on_stack();
+}
 }
 }
 };
 static MarkActivationClosure mark_activation_closure;
+class SetHotnessClosure: public CodeBlobClosure {
+public:
+virtual void do_code_blob(CodeBlob* cb) {
+if (cb->is_nmethod()) {
+nmethod* nm = (nmethod*)cb;
+nm->set_hotness_counter(NMethodSweeper::hotness_counter_reset_val());
+}
+}
+};
+static SetHotnessClosure set_hotness_closure;
+int NMethodSweeper::hotness_counter_reset_val() {
+if (_hotness_counter_reset_val == 0) {
+_hotness_counter_reset_val = (ReservedCodeCacheSize < M) ? 1 : (ReservedCodeCacheSize / M) * 2;
+}
+return _hotness_counter_reset_val;
+}
 bool NMethodSweeper::sweep_in_progress() {
 return (_current != NULL);
 }
-void NMethodSweeper::scan_stacks() {
+// Scans the stacks of all Java threads and marks activations of not-entrant methods.
+// No need to synchronize access, since 'mark_active_nmethods' is always executed at a
+// safepoint.
+void NMethodSweeper::mark_active_nmethods() {
 assert(SafepointSynchronize::is_at_safepoint(), "must be executed at a safepoint");
-if (!MethodFlushing) return;
+// If we do not want to reclaim not-entrant or zombie methods there is no need
+// to scan stacks
-// No need to synchronize access, since this is always executed at a
+if (!MethodFlushing) {
-// safepoint.
+return;
+}
-// Make sure CompiledIC_lock in unlocked, since we might update some
-// inline caches. If it is, we just bail-out and try later.
-if (CompiledIC_lock->is_locked() || Patching_lock->is_locked()) return;
 // Check for restart
 assert(CodeCache::find_blob_unsafe(_current) == _current, "Sweeper nmethod cached state invalid");
-if (!sweep_in_progress() && _resweep) {
+if (!sweep_in_progress() && need_marking_phase()) {
 _seen        = 0;
 _invocations = NmethodSweepFraction;
 _current     = CodeCache::first_nmethod();
 _traversals  += 1;
 _total_time_this_sweep = 0;
 tty->print_cr("### Sweep: stack traversal %d", _traversals);
 }
 Threads::nmethods_do(&mark_activation_closure);
 // reset the flags since we started a scan from the beginning.
-_resweep = false;
+reset_nmethod_marking();
 _locked_seen = 0;
 _not_entrant_seen_on_stack = 0;
-}
+} else {
+// Only set hotness counter
-if (UseCodeCacheFlushing) {
+Threads::nmethods_do(&set_hotness_closure);
-// only allow new flushes after the interval is complete.
+}
-jlong now           = os::javaTimeMillis();
-jlong max_interval  = (jlong)MinCodeCacheFlushingInterval * (jlong)1000;
+OrderAccess::storestore();
-jlong curr_interval = now - _last_full_flush_time;
-if (curr_interval > max_interval) {
-_flush_token = 0;
-}
-if (!CodeCache::needs_flushing() && !CompileBroker::should_compile_new_jobs()) {
-CompileBroker::set_should_compile_new_jobs(CompileBroker::run_compilation);
-log_sweep("restart_compiler");
-}
-}
 }
 void NMethodSweeper::possibly_sweep() {
 assert(JavaThread::current()->thread_state() == _thread_in_vm, "must run in vm mode");
-if (!MethodFlushing || !sweep_in_progress()) return;
+if (!MethodFlushing || !sweep_in_progress()) {
+return;
+}
 if (_invocations > 0) {
 // Only one thread at a time will sweep
 jint old = Atomic::cmpxchg( 1, &_sweep_started, 0 );
 if (old != 0) {
 }
 if (!CompileBroker::should_compile_new_jobs()) {
 // If we have turned off compilations we might as well do full sweeps
 // in order to reach the clean state faster. Otherwise the sleeping compiler
-// threads will slow down sweeping. After a few iterations the cache
+// threads will slow down sweeping.
-// will be clean and sweeping stops (_resweep will not be set)
 _invocations = 1;
 }
 // We want to visit all nmethods after NmethodSweepFraction
 // invocations so divide the remaining number of nmethods by the
 // the number of nmethods changes during the sweep so the final
 // stage must iterate until it there are no more nmethods.
 int todo = (CodeCache::nof_nmethods() - _seen) / _invocations;
 int swept_count = 0;
 assert(!SafepointSynchronize::is_at_safepoint(), "should not be in safepoint when we get here");
 assert(!CodeCache_lock->owned_by_self(), "just checking");
+int freed_memory = 0;
 {
 MutexLockerEx mu(CodeCache_lock, Mutex::_no_safepoint_check_flag);
 // The last invocation iterates until there are no more nmethods
 for (int i = 0; (i < todo || _invocations == 1) && _current != NULL; i++) {
 nmethod* next = CodeCache::next_nmethod(_current);
 // Now ready to process nmethod and give up CodeCache_lock
 {
 MutexUnlockerEx mu(CodeCache_lock, Mutex::_no_safepoint_check_flag);
-process_nmethod(_current);
+freed_memory += process_nmethod(_current);
 }
 _seen++;
 _current = next;
 }
 }
 assert(_invocations > 1 || _current == NULL, "must have scanned the whole cache");
-if (!sweep_in_progress() && !_resweep && (_locked_seen || _not_entrant_seen_on_stack)) {
+if (!sweep_in_progress() && !need_marking_phase() && (_locked_seen || _not_entrant_seen_on_stack)) {
 // we've completed a scan without making progress but there were
 // nmethods we were unable to process either because they were
-// locked or were still on stack.  We don't have to aggresively
+// locked or were still on stack. We don't have to aggressively
-// clean them up so just stop scanning.  We could scan once more
+// clean them up so just stop scanning. We could scan once more
 // but that complicates the control logic and it's unlikely to
 // matter much.
 if (PrintMethodFlushing) {
 tty->print_cr("### Couldn't make progress on some nmethods so stopping sweep");
 }
 if (_invocations == 1) {
 _peak_sweep_time = MAX2(_peak_sweep_time, _total_time_this_sweep);
 log_sweep("finished");
 }
-// Sweeper is the only case where memory is released,
+// Sweeper is the only case where memory is released, check here if it
-// check here if it is time to restart the compiler.
+// is time to restart the compiler. Only checking if there is a certain
-if (UseCodeCacheFlushing && !CompileBroker::should_compile_new_jobs() && !CodeCache::needs_flushing()) {
+// amount of free memory in the code cache might lead to re-enabling
+// compilation although no memory has been released. For example, there are
+// cases when compilation was disabled although there is 4MB (or more) free
+// memory in the code cache. The reason is code cache fragmentation. Therefore,
+// it only makes sense to re-enable compilation if we have actually freed memory.
+// Note that typically several kB are released for sweeping 16MB of the code
+// cache. As a result, 'freed_memory' > 0 to restart the compiler.
+if (UseCodeCacheFlushing && (!CompileBroker::should_compile_new_jobs() && (freed_memory > 0))) {
 CompileBroker::set_should_compile_new_jobs(CompileBroker::run_compilation);
 log_sweep("restart_compiler");
 }
 }
 public:
 NMethodMarker(nmethod* nm) {
 _thread = CompilerThread::current();
 if (!nm->is_zombie() && !nm->is_unloaded()) {
 // Only expose live nmethods for scanning
 _thread->set_scanned_nmethod(nm);
 }
 }
 ~NMethodMarker() {
 _thread->set_scanned_nmethod(NULL);
 }
 };
 MutexLockerEx mu(CodeCache_lock, Mutex::_no_safepoint_check_flag);
 nm->flush();
 }
-void NMethodSweeper::process_nmethod(nmethod *nm) {
+int NMethodSweeper::process_nmethod(nmethod *nm) {
 assert(!CodeCache_lock->owned_by_self(), "just checking");
+int freed_memory = 0;
 // Make sure this nmethod doesn't get unloaded during the scan,
-// since the locks acquired below might safepoint.
+// since safepoints may happen during acquired below locks.
 NMethodMarker nmm(nm);
 SWEEP(nm);
 // Skip methods that are currently referenced by the VM
 if (nm->is_locked_by_vm()) {
 // But still remember to clean-up inline caches for alive nmethods
 if (nm->is_alive()) {
-// Clean-up all inline caches that points to zombie/non-reentrant methods
+// Clean inline caches that point to zombie/non-entrant methods
 MutexLocker cl(CompiledIC_lock);
 nm->cleanup_inline_caches();
 SWEEP(nm);
 } else {
 _locked_seen++;
 SWEEP(nm);
 }
-return;
+return freed_memory;
 }
 if (nm->is_zombie()) {
-// If it is first time, we see nmethod then we mark it. Otherwise,
+// If it is the first time we see nmethod then we mark it. Otherwise,
-// we reclame it. When we have seen a zombie method twice, we know that
+// we reclaim it. When we have seen a zombie method twice, we know that
 // there are no inline caches that refer to it.
 if (nm->is_marked_for_reclamation()) {
 assert(!nm->is_locked_by_vm(), "must not flush locked nmethods");
 if (PrintMethodFlushing && Verbose) {
 tty->print_cr("### Nmethod %3d/" PTR_FORMAT " (marked for reclamation) being flushed", nm->compile_id(), nm);
 }
+freed_memory = nm->total_size();
 release_nmethod(nm);
 _flushed_count++;
 } else {
 if (PrintMethodFlushing && Verbose) {
 tty->print_cr("### Nmethod %3d/" PTR_FORMAT " (zombie) being marked for reclamation", nm->compile_id(), nm);
 }
 nm->mark_for_reclamation();
-_resweep = true;
+request_nmethod_marking();
 _marked_count++;
 SWEEP(nm);
 }
 } else if (nm->is_not_entrant()) {
-// If there is no current activations of this method on the
+// If there are no current activations of this method on the
 // stack we can safely convert it to a zombie method
 if (nm->can_not_entrant_be_converted()) {
 if (PrintMethodFlushing && Verbose) {
 tty->print_cr("### Nmethod %3d/" PTR_FORMAT " (not entrant) being made zombie", nm->compile_id(), nm);
 }
 nm->make_zombie();
-_resweep = true;
+request_nmethod_marking();
 _zombified_count++;
 SWEEP(nm);
 } else {
 // Still alive, clean up its inline caches
 MutexLocker cl(CompiledIC_lock);
 _not_entrant_seen_on_stack++;
 SWEEP(nm);
 }
 } else if (nm->is_unloaded()) {
 // Unloaded code, just make it a zombie
-if (PrintMethodFlushing && Verbose)
+if (PrintMethodFlushing && Verbose) {
 tty->print_cr("### Nmethod %3d/" PTR_FORMAT " (unloaded) being made zombie", nm->compile_id(), nm);
+}
 if (nm->is_osr_method()) {
 SWEEP(nm);
 // No inline caches will ever point to osr methods, so we can just remove it
+freed_memory = nm->total_size();
 release_nmethod(nm);
 _flushed_count++;
 } else {
 nm->make_zombie();
-_resweep = true;
+request_nmethod_marking();
 _zombified_count++;
 SWEEP(nm);
 }
 } else {
-assert(nm->is_alive(), "should be alive");
 if (UseCodeCacheFlushing) {
-if (nm->is_speculatively_disconnected() && !nm->is_locked_by_vm() && !nm->is_osr_method() &&
+if (!nm->is_locked_by_vm() && !nm->is_osr_method() && !nm->is_native_method()) {
-(_traversals > _last_flush_traversal_id + 2) && (nm->compile_id() < _highest_marked)) {
+// Do not make native methods and OSR-methods not-entrant
-// This method has not been called since the forced cleanup happened
+nm->dec_hotness_counter();
-nm->make_not_entrant();
+// Get the initial value of the hotness counter. This value depends on the
-}
+// ReservedCodeCacheSize
-}
+int reset_val = hotness_counter_reset_val();
+int time_since_reset = reset_val - nm->hotness_counter();
-// Clean-up all inline caches that points to zombie/non-reentrant methods
+double threshold = -reset_val + (CodeCache::reverse_free_ratio() * NmethodSweepActivity);
+// The less free space in the code cache we have - the bigger reverse_free_ratio() is.
+// I.e., 'threshold' increases with lower available space in the code cache and a higher
+// NmethodSweepActivity. If the current hotness counter - which decreases from its initial
+// value until it is reset by stack walking - is smaller than the computed threshold, the
+// corresponding nmethod is considered for removal.
+if ((NmethodSweepActivity > 0) && (nm->hotness_counter() < threshold) && (time_since_reset > 10)) {
+// A method is marked as not-entrant if the method is
+// 1) 'old enough': nm->hotness_counter() < threshold
+// 2) The method was in_use for a minimum amount of time: (time_since_reset > 10)
+//    The second condition is necessary if we are dealing with very small code cache
+//    sizes (e.g., <10m) and the code cache size is too small to hold all hot methods.
+//    The second condition ensures that methods are not immediately made not-entrant
+//    after compilation.
+nm->make_not_entrant();
+request_nmethod_marking();
+}
+}
+}
+// Clean-up all inline caches that point to zombie/non-reentrant methods
 MutexLocker cl(CompiledIC_lock);
 nm->cleanup_inline_caches();
 SWEEP(nm);
 }
-}
+return freed_memory;
+}
-// Code cache unloading: when compilers notice the code cache is getting full,
-// they will call a vm op that comes here. This code attempts to speculatively
-// unload the oldest half of the nmethods (based on the compile job id) by
-// saving the old code in a list in the CodeCache. Then
-// execution resumes. If a method so marked is not called by the second sweeper
-// stack traversal after the current one, the nmethod will be marked non-entrant and
-// got rid of by normal sweeping. If the method is called, the Method*'s
-// _code field is restored and the Method*/nmethod
-// go back to their normal state.
-void NMethodSweeper::handle_full_code_cache(bool is_full) {
-if (is_full) {
-// Since code cache is full, immediately stop new compiles
-if (CompileBroker::set_should_compile_new_jobs(CompileBroker::stop_compilation)) {
-log_sweep("disable_compiler");
-}
-}
-// Make sure only one thread can flush
-// The token is reset after CodeCacheMinimumFlushInterval in scan stacks,
-// no need to check the timeout here.
-jint old = Atomic::cmpxchg( 1, &_flush_token, 0 );
-if (old != 0) {
-return;
-}
-VM_HandleFullCodeCache op(is_full);
-VMThread::execute(&op);
-// resweep again as soon as possible
-_resweep = true;
-}
-void NMethodSweeper::speculative_disconnect_nmethods(bool is_full) {
-// If there was a race in detecting full code cache, only run
-// one vm op for it or keep the compiler shut off
-jlong disconnect_start_counter = os::elapsed_counter();
-// Traverse the code cache trying to dump the oldest nmethods
-int curr_max_comp_id = CompileBroker::get_compilation_id();
-int flush_target = ((curr_max_comp_id - _dead_compile_ids) / CodeCacheFlushingFraction) + _dead_compile_ids;
-log_sweep("start_cleaning");
-nmethod* nm = CodeCache::alive_nmethod(CodeCache::first());
-jint disconnected = 0;
-jint made_not_entrant  = 0;
-jint nmethod_count = 0;
-while ((nm != NULL)){
-int curr_comp_id = nm->compile_id();
-// OSR methods cannot be flushed like this. Also, don't flush native methods
-// since they are part of the JDK in most cases
-if (!nm->is_osr_method() && !nm->is_locked_by_vm() && !nm->is_native_method()) {
-// only count methods that can be speculatively disconnected
-nmethod_count++;
-if (nm->is_in_use() && (curr_comp_id < flush_target)) {
-if ((nm->method()->code() == nm)) {
-// This method has not been previously considered for
-// unloading or it was restored already
-CodeCache::speculatively_disconnect(nm);
-disconnected++;
-} else if (nm->is_speculatively_disconnected()) {
-// This method was previously considered for preemptive unloading and was not called since then
-CompilationPolicy::policy()->delay_compilation(nm->method());
-nm->make_not_entrant();
-made_not_entrant++;
-}
-if (curr_comp_id > _highest_marked) {
-_highest_marked = curr_comp_id;
-}
-}
-}
-nm = CodeCache::alive_nmethod(CodeCache::next(nm));
-}
-// remember how many compile_ids wheren't seen last flush.
-_dead_compile_ids = curr_max_comp_id - nmethod_count;
-log_sweep("stop_cleaning",
-"disconnected='" UINT32_FORMAT "' made_not_entrant='" UINT32_FORMAT "'",
-disconnected, made_not_entrant);
-// Shut off compiler. Sweeper will start over with a new stack scan and
-// traversal cycle and turn it back on if it clears enough space.
-if (is_full) {
-_last_full_flush_time = os::javaTimeMillis();
-}
-jlong disconnect_end_counter = os::elapsed_counter();
-jlong disconnect_time = disconnect_end_counter - disconnect_start_counter;
-_total_disconnect_time += disconnect_time;
-_peak_disconnect_time = MAX2(disconnect_time, _peak_disconnect_time);
-EventCleanCodeCache event(UNTIMED);
-if (event.should_commit()) {
-event.set_starttime(disconnect_start_counter);
-event.set_endtime(disconnect_end_counter);
-event.set_disconnectedCount(disconnected);
-event.set_madeNonEntrantCount(made_not_entrant);
-event.commit();
-}
-_number_of_flushes++;
-// After two more traversals the sweeper will get rid of unrestored nmethods
-_last_flush_traversal_id = _traversals;
-_resweep = true;
-#ifdef ASSERT
-if(PrintMethodFlushing && Verbose) {
-tty->print_cr("### sweeper: unload time: " INT64_FORMAT, (jlong)disconnect_time);
-}
-#endif
-}
 // Print out some state information about the current sweep and the
 // state of the code cache if it's requested.
 void NMethodSweeper::log_sweep(const char* msg, const char* format, ...) {
 if (PrintMethodFlushing) {

Mercurial > hg > graal-jvmci-8

comparison src/share/vm/runtime/sweeper.cpp @ 12324:510fbd28919c