src/hotspot/share/gc/g1/g1RegionToSpaceMapper.cpp - platform/libcore - Git at Google

 /*
  * Copyright (c) 2001, 2019, Oracle and/or its affiliates. All rights reserved.
  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
  *
  * This code is free software; you can redistribute it and/or modify it
  * under the terms of the GNU General Public License version 2 only, as
  * published by the Free Software Foundation.
  *
  * This code is distributed in the hope that it will be useful, but WITHOUT
  * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
  * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
  * version 2 for more details (a copy is included in the LICENSE file that
  * accompanied this code).
  *
  * You should have received a copy of the GNU General Public License version
  * 2 along with this work; if not, write to the Free Software Foundation,
  * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
  *
  * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
  * or visit www.oracle.com if you need additional information or have any
  * questions.
  *
  */

 #include "precompiled.hpp"
 #include "gc/g1/g1BiasedArray.hpp"
 #include "gc/g1/g1NUMA.hpp"
 #include "gc/g1/g1RegionToSpaceMapper.hpp"
 #include "memory/allocation.inline.hpp"
 #include "memory/virtualspace.hpp"
 #include "runtime/mutexLocker.hpp"
 #include "services/memTracker.hpp"
 #include "utilities/align.hpp"
 #include "utilities/bitMap.inline.hpp"
 #include "utilities/powerOfTwo.hpp"

 G1RegionToSpaceMapper::G1RegionToSpaceMapper(ReservedSpace rs,
                                              size_t used_size,
                                              size_t page_size,
                                              size_t region_granularity,
                                              size_t commit_factor,
                                              MEMFLAGS type) :
   _listener(NULL),
   _storage(rs, used_size, page_size),
   _region_granularity(region_granularity),
   _region_commit_map(rs.size() * commit_factor / region_granularity, mtGC),
   _memory_type(type) {
   guarantee(is_power_of_2(page_size), "must be");
   guarantee(is_power_of_2(region_granularity), "must be");

   MemTracker::record_virtual_memory_type((address)rs.base(), type);
 }

 // Used to manually signal a mapper to handle a set of regions as committed.
 // Setting the 'zero_filled' parameter to false signals the mapper that the
 // regions have not been cleared by the OS and that they need to be clear
 // explicitly.
 void G1RegionToSpaceMapper::signal_mapping_changed(uint start_idx, size_t num_regions) {
   fire_on_commit(start_idx, num_regions, false);
 }

 // G1RegionToSpaceMapper implementation where the region granularity is larger than
 // or the same as the commit granularity.
 // Basically, the space corresponding to one region region spans several OS pages.
 class G1RegionsLargerThanCommitSizeMapper : public G1RegionToSpaceMapper {
  private:
   size_t _pages_per_region;

  public:
   G1RegionsLargerThanCommitSizeMapper(ReservedSpace rs,
                                       size_t actual_size,
                                       size_t page_size,
                                       size_t alloc_granularity,
                                       size_t commit_factor,
                                       MEMFLAGS type) :
     G1RegionToSpaceMapper(rs, actual_size, page_size, alloc_granularity, commit_factor, type),
     _pages_per_region(alloc_granularity / (page_size * commit_factor)) {

     guarantee(alloc_granularity >= page_size, "allocation granularity smaller than commit granularity");
   }

   bool is_range_committed(uint start_idx, size_t num_regions) {
     BitMap::idx_t end = start_idx + num_regions;
     return _region_commit_map.get_next_zero_offset(start_idx, end) == end;
   }

   bool is_range_uncommitted(uint start_idx, size_t num_regions) {
     BitMap::idx_t end = start_idx + num_regions;
     return _region_commit_map.get_next_one_offset(start_idx, end) == end;
   }

   virtual void commit_regions(uint start_idx, size_t num_regions, WorkGang* pretouch_gang) {
     guarantee(is_range_uncommitted(start_idx, num_regions),
               "Range not uncommitted, start: %u, num_regions: " SIZE_FORMAT,
               start_idx, num_regions);

     const size_t start_page = (size_t)start_idx * _pages_per_region;
     const size_t size_in_pages = num_regions * _pages_per_region;
     bool zero_filled = _storage.commit(start_page, size_in_pages);
     if (_memory_type == mtJavaHeap) {
       for (uint region_index = start_idx; region_index < start_idx + num_regions; region_index++ ) {
         void* address = _storage.page_start(region_index * _pages_per_region);
         size_t size_in_bytes = _storage.page_size() * _pages_per_region;
         G1NUMA::numa()->request_memory_on_node(address, size_in_bytes, region_index);
       }
     }
     if (AlwaysPreTouch) {
       _storage.pretouch(start_page, size_in_pages, pretouch_gang);
     }
     _region_commit_map.par_set_range(start_idx, start_idx + num_regions, BitMap::unknown_range);
     fire_on_commit(start_idx, num_regions, zero_filled);
   }

   virtual void uncommit_regions(uint start_idx, size_t num_regions) {
     guarantee(is_range_committed(start_idx, num_regions),
              "Range not committed, start: %u, num_regions: " SIZE_FORMAT,
               start_idx, num_regions);

     _storage.uncommit((size_t)start_idx * _pages_per_region, num_regions * _pages_per_region);
     _region_commit_map.par_clear_range(start_idx, start_idx + num_regions, BitMap::unknown_range);
   }
 };

 // G1RegionToSpaceMapper implementation where the region granularity is smaller
 // than the commit granularity.
 // Basically, the contents of one OS page span several regions.
 class G1RegionsSmallerThanCommitSizeMapper : public G1RegionToSpaceMapper {
   size_t _regions_per_page;
   // Lock to prevent bitmap updates and the actual underlying
   // commit to get out of order. This can happen in the cases
   // where one thread is expanding the heap during a humongous
   // allocation and at the same time the service thread is
   // doing uncommit. These operations will not operate on the
   // same regions, but they might operate on regions sharing
   // an underlying OS page. So we need to make sure that both
   // those resources are in sync:
   // - G1RegionToSpaceMapper::_region_commit_map;
   // - G1PageBasedVirtualSpace::_committed (_storage.commit())
   Mutex _lock;

   size_t region_idx_to_page_idx(uint region_idx) const {
     return region_idx / _regions_per_page;
   }

   bool is_page_committed(size_t page_idx) {
     size_t region = page_idx * _regions_per_page;
     size_t region_limit = region + _regions_per_page;
     // Committed if there is a bit set in the range.
     return _region_commit_map.get_next_one_offset(region, region_limit) != region_limit;
   }

   void numa_request_on_node(size_t page_idx) {
     if (_memory_type == mtJavaHeap) {
       uint region = (uint)(page_idx * _regions_per_page);
       void* address = _storage.page_start(page_idx);
       size_t size_in_bytes = _storage.page_size();
       G1NUMA::numa()->request_memory_on_node(address, size_in_bytes, region);
     }
   }

  public:
   G1RegionsSmallerThanCommitSizeMapper(ReservedSpace rs,
                                        size_t actual_size,
                                        size_t page_size,
                                        size_t alloc_granularity,
                                        size_t commit_factor,
                                        MEMFLAGS type) :
     G1RegionToSpaceMapper(rs, actual_size, page_size, alloc_granularity, commit_factor, type),
     _regions_per_page((page_size * commit_factor) / alloc_granularity),
     _lock(Mutex::leaf, "G1 mapper lock", true, Mutex::_safepoint_check_never) {

     guarantee((page_size * commit_factor) >= alloc_granularity, "allocation granularity smaller than commit granularity");
   }

   virtual void commit_regions(uint start_idx, size_t num_regions, WorkGang* pretouch_gang) {
     uint region_limit = (uint)(start_idx + num_regions);
     assert(num_regions > 0, "Must commit at least one region");
     assert(_region_commit_map.get_next_one_offset(start_idx, region_limit) == region_limit,
            "Should be no committed regions in the range [%u, %u)", start_idx, region_limit);

     size_t const NoPage = ~(size_t)0;

     size_t first_committed = NoPage;
     size_t num_committed = 0;

     size_t start_page = region_idx_to_page_idx(start_idx);
     size_t end_page = region_idx_to_page_idx(region_limit - 1);

     bool all_zero_filled = true;

     // Concurrent operations might operate on regions sharing the same
     // underlying OS page. See lock declaration for more details.
     {
       MutexLocker ml(&_lock, Mutex::_no_safepoint_check_flag);
       for (size_t page = start_page; page <= end_page; page++) {
         if (!is_page_committed(page)) {
           // Page not committed.
           if (num_committed == 0) {
             first_committed = page;
           }
           num_committed++;

           if (!_storage.commit(page, 1)) {
             // Found dirty region during commit.
             all_zero_filled = false;
           }

           // Move memory to correct NUMA node for the heap.
           numa_request_on_node(page);
         } else {
           // Page already committed.
           all_zero_filled = false;
         }
       }

       // Update the commit map for the given range. Not using the par_set_range
       // since updates to _region_commit_map for this mapper is protected by _lock.
       _region_commit_map.set_range(start_idx, region_limit, BitMap::unknown_range);
     }

     if (AlwaysPreTouch && num_committed > 0) {
       _storage.pretouch(first_committed, num_committed, pretouch_gang);
     }

     fire_on_commit(start_idx, num_regions, all_zero_filled);
   }

   virtual void uncommit_regions(uint start_idx, size_t num_regions) {
     uint region_limit = (uint)(start_idx + num_regions);
     assert(num_regions > 0, "Must uncommit at least one region");
     assert(_region_commit_map.get_next_zero_offset(start_idx, region_limit) == region_limit,
            "Should only be committed regions in the range [%u, %u)", start_idx, region_limit);

     size_t start_page = region_idx_to_page_idx(start_idx);
     size_t end_page = region_idx_to_page_idx(region_limit - 1);

     // Concurrent operations might operate on regions sharing the same
     // underlying OS page. See lock declaration for more details.
     MutexLocker ml(&_lock, Mutex::_no_safepoint_check_flag);
     // Clear commit map for the given range. Not using the par_clear_range since
     // updates to _region_commit_map for this mapper is protected by _lock.
     _region_commit_map.clear_range(start_idx, region_limit, BitMap::unknown_range);

     for (size_t page = start_page; page <= end_page; page++) {
       // We know all pages were committed before clearing the map. If the
       // the page is still marked as committed after the clear we should
       // not uncommit it.
       if (!is_page_committed(page)) {
         _storage.uncommit(page, 1);
       }
     }
   }
 };

 void G1RegionToSpaceMapper::fire_on_commit(uint start_idx, size_t num_regions, bool zero_filled) {
   if (_listener != NULL) {
     _listener->on_commit(start_idx, num_regions, zero_filled);
   }
 }

 G1RegionToSpaceMapper* G1RegionToSpaceMapper::create_mapper(ReservedSpace rs,
                                                             size_t actual_size,
                                                             size_t page_size,
                                                             size_t region_granularity,
                                                             size_t commit_factor,
                                                             MEMFLAGS type) {
   if (region_granularity >= (page_size * commit_factor)) {
     return new G1RegionsLargerThanCommitSizeMapper(rs, actual_size, page_size, region_granularity, commit_factor, type);
   } else {
     return new G1RegionsSmallerThanCommitSizeMapper(rs, actual_size, page_size, region_granularity, commit_factor, type);
   }
 }
	/*
	* Copyright (c) 2001, 2019, Oracle and/or its affiliates. All rights reserved.
	* DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
	*
	* This code is free software; you can redistribute it and/or modify it
	* under the terms of the GNU General Public License version 2 only, as
	* published by the Free Software Foundation.
	*
	* This code is distributed in the hope that it will be useful, but WITHOUT
	* ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
	* FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
	* version 2 for more details (a copy is included in the LICENSE file that
	* accompanied this code).
	*
	* You should have received a copy of the GNU General Public License version
	* 2 along with this work; if not, write to the Free Software Foundation,
	* Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
	*
	* Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
	* or visit www.oracle.com if you need additional information or have any
	* questions.
	*
	*/

	#include "precompiled.hpp"
	#include "gc/g1/g1BiasedArray.hpp"
	#include "gc/g1/g1NUMA.hpp"
	#include "gc/g1/g1RegionToSpaceMapper.hpp"
	#include "memory/allocation.inline.hpp"
	#include "memory/virtualspace.hpp"
	#include "runtime/mutexLocker.hpp"
	#include "services/memTracker.hpp"
	#include "utilities/align.hpp"
	#include "utilities/bitMap.inline.hpp"
	#include "utilities/powerOfTwo.hpp"

	G1RegionToSpaceMapper::G1RegionToSpaceMapper(ReservedSpace rs,
	size_t used_size,
	size_t page_size,
	size_t region_granularity,
	size_t commit_factor,
	MEMFLAGS type) :
	_listener(NULL),
	_storage(rs, used_size, page_size),
	_region_granularity(region_granularity),
	_region_commit_map(rs.size() * commit_factor / region_granularity, mtGC),
	_memory_type(type) {
	guarantee(is_power_of_2(page_size), "must be");
	guarantee(is_power_of_2(region_granularity), "must be");

	MemTracker::record_virtual_memory_type((address)rs.base(), type);
	}

	// Used to manually signal a mapper to handle a set of regions as committed.
	// Setting the 'zero_filled' parameter to false signals the mapper that the
	// regions have not been cleared by the OS and that they need to be clear
	// explicitly.
	void G1RegionToSpaceMapper::signal_mapping_changed(uint start_idx, size_t num_regions) {
	fire_on_commit(start_idx, num_regions, false);
	}

	// G1RegionToSpaceMapper implementation where the region granularity is larger than
	// or the same as the commit granularity.
	// Basically, the space corresponding to one region region spans several OS pages.
	class G1RegionsLargerThanCommitSizeMapper : public G1RegionToSpaceMapper {
	private:
	size_t _pages_per_region;

	public:
	G1RegionsLargerThanCommitSizeMapper(ReservedSpace rs,
	size_t actual_size,
	size_t page_size,
	size_t alloc_granularity,
	size_t commit_factor,
	MEMFLAGS type) :
	G1RegionToSpaceMapper(rs, actual_size, page_size, alloc_granularity, commit_factor, type),
	_pages_per_region(alloc_granularity / (page_size * commit_factor)) {

	guarantee(alloc_granularity >= page_size, "allocation granularity smaller than commit granularity");
	}

	bool is_range_committed(uint start_idx, size_t num_regions) {
	BitMap::idx_t end = start_idx + num_regions;
	return _region_commit_map.get_next_zero_offset(start_idx, end) == end;
	}

	bool is_range_uncommitted(uint start_idx, size_t num_regions) {
	BitMap::idx_t end = start_idx + num_regions;
	return _region_commit_map.get_next_one_offset(start_idx, end) == end;
	}

	virtual void commit_regions(uint start_idx, size_t num_regions, WorkGang* pretouch_gang) {
	guarantee(is_range_uncommitted(start_idx, num_regions),
	"Range not uncommitted, start: %u, num_regions: " SIZE_FORMAT,
	start_idx, num_regions);

	const size_t start_page = (size_t)start_idx * _pages_per_region;
	const size_t size_in_pages = num_regions * _pages_per_region;
	bool zero_filled = _storage.commit(start_page, size_in_pages);
	if (_memory_type == mtJavaHeap) {
	for (uint region_index = start_idx; region_index < start_idx + num_regions; region_index++ ) {
	void* address = _storage.page_start(region_index * _pages_per_region);
	size_t size_in_bytes = _storage.page_size() * _pages_per_region;
	G1NUMA::numa()->request_memory_on_node(address, size_in_bytes, region_index);
	}
	}
	if (AlwaysPreTouch) {
	_storage.pretouch(start_page, size_in_pages, pretouch_gang);
	}
	_region_commit_map.par_set_range(start_idx, start_idx + num_regions, BitMap::unknown_range);
	fire_on_commit(start_idx, num_regions, zero_filled);
	}

	virtual void uncommit_regions(uint start_idx, size_t num_regions) {
	guarantee(is_range_committed(start_idx, num_regions),
	"Range not committed, start: %u, num_regions: " SIZE_FORMAT,
	start_idx, num_regions);

	_storage.uncommit((size_t)start_idx * _pages_per_region, num_regions * _pages_per_region);
	_region_commit_map.par_clear_range(start_idx, start_idx + num_regions, BitMap::unknown_range);
	}
	};

	// G1RegionToSpaceMapper implementation where the region granularity is smaller
	// than the commit granularity.
	// Basically, the contents of one OS page span several regions.
	class G1RegionsSmallerThanCommitSizeMapper : public G1RegionToSpaceMapper {
	size_t _regions_per_page;
	// Lock to prevent bitmap updates and the actual underlying
	// commit to get out of order. This can happen in the cases
	// where one thread is expanding the heap during a humongous
	// allocation and at the same time the service thread is
	// doing uncommit. These operations will not operate on the
	// same regions, but they might operate on regions sharing
	// an underlying OS page. So we need to make sure that both
	// those resources are in sync:
	// - G1RegionToSpaceMapper::_region_commit_map;
	// - G1PageBasedVirtualSpace::_committed (_storage.commit())
	Mutex _lock;

	size_t region_idx_to_page_idx(uint region_idx) const {
	return region_idx / _regions_per_page;
	}

	bool is_page_committed(size_t page_idx) {
	size_t region = page_idx * _regions_per_page;
	size_t region_limit = region + _regions_per_page;
	// Committed if there is a bit set in the range.
	return _region_commit_map.get_next_one_offset(region, region_limit) != region_limit;
	}

	void numa_request_on_node(size_t page_idx) {
	if (_memory_type == mtJavaHeap) {
	uint region = (uint)(page_idx * _regions_per_page);
	void* address = _storage.page_start(page_idx);
	size_t size_in_bytes = _storage.page_size();
	G1NUMA::numa()->request_memory_on_node(address, size_in_bytes, region);
	}
	}

	public:
	G1RegionsSmallerThanCommitSizeMapper(ReservedSpace rs,
	size_t actual_size,
	size_t page_size,
	size_t alloc_granularity,
	size_t commit_factor,
	MEMFLAGS type) :
	G1RegionToSpaceMapper(rs, actual_size, page_size, alloc_granularity, commit_factor, type),
	_regions_per_page((page_size * commit_factor) / alloc_granularity),
	_lock(Mutex::leaf, "G1 mapper lock", true, Mutex::_safepoint_check_never) {

	guarantee((page_size * commit_factor) >= alloc_granularity, "allocation granularity smaller than commit granularity");
	}

	virtual void commit_regions(uint start_idx, size_t num_regions, WorkGang* pretouch_gang) {
	uint region_limit = (uint)(start_idx + num_regions);
	assert(num_regions > 0, "Must commit at least one region");
	assert(_region_commit_map.get_next_one_offset(start_idx, region_limit) == region_limit,
	"Should be no committed regions in the range [%u, %u)", start_idx, region_limit);

	size_t const NoPage = ~(size_t)0;

	size_t first_committed = NoPage;
	size_t num_committed = 0;

	size_t start_page = region_idx_to_page_idx(start_idx);
	size_t end_page = region_idx_to_page_idx(region_limit - 1);

	bool all_zero_filled = true;

	// Concurrent operations might operate on regions sharing the same
	// underlying OS page. See lock declaration for more details.
	{
	MutexLocker ml(&_lock, Mutex::_no_safepoint_check_flag);
	for (size_t page = start_page; page <= end_page; page++) {
	if (!is_page_committed(page)) {
	// Page not committed.
	if (num_committed == 0) {
	first_committed = page;
	}
	num_committed++;

	if (!_storage.commit(page, 1)) {
	// Found dirty region during commit.
	all_zero_filled = false;
	}

	// Move memory to correct NUMA node for the heap.
	numa_request_on_node(page);
	} else {
	// Page already committed.
	all_zero_filled = false;
	}
	}

	// Update the commit map for the given range. Not using the par_set_range
	// since updates to _region_commit_map for this mapper is protected by _lock.
	_region_commit_map.set_range(start_idx, region_limit, BitMap::unknown_range);
	}

	if (AlwaysPreTouch && num_committed > 0) {
	_storage.pretouch(first_committed, num_committed, pretouch_gang);
	}

	fire_on_commit(start_idx, num_regions, all_zero_filled);
	}

	virtual void uncommit_regions(uint start_idx, size_t num_regions) {
	uint region_limit = (uint)(start_idx + num_regions);
	assert(num_regions > 0, "Must uncommit at least one region");
	assert(_region_commit_map.get_next_zero_offset(start_idx, region_limit) == region_limit,
	"Should only be committed regions in the range [%u, %u)", start_idx, region_limit);

	size_t start_page = region_idx_to_page_idx(start_idx);
	size_t end_page = region_idx_to_page_idx(region_limit - 1);

	// Concurrent operations might operate on regions sharing the same
	// underlying OS page. See lock declaration for more details.
	MutexLocker ml(&_lock, Mutex::_no_safepoint_check_flag);
	// Clear commit map for the given range. Not using the par_clear_range since
	// updates to _region_commit_map for this mapper is protected by _lock.
	_region_commit_map.clear_range(start_idx, region_limit, BitMap::unknown_range);

	for (size_t page = start_page; page <= end_page; page++) {
	// We know all pages were committed before clearing the map. If the
	// the page is still marked as committed after the clear we should
	// not uncommit it.
	if (!is_page_committed(page)) {
	_storage.uncommit(page, 1);
	}
	}
	}
	};

	void G1RegionToSpaceMapper::fire_on_commit(uint start_idx, size_t num_regions, bool zero_filled) {
	if (_listener != NULL) {
	_listener->on_commit(start_idx, num_regions, zero_filled);
	}
	}

	G1RegionToSpaceMapper* G1RegionToSpaceMapper::create_mapper(ReservedSpace rs,
	size_t actual_size,
	size_t page_size,
	size_t region_granularity,
	size_t commit_factor,
	MEMFLAGS type) {
	if (region_granularity >= (page_size * commit_factor)) {
	return new G1RegionsLargerThanCommitSizeMapper(rs, actual_size, page_size, region_granularity, commit_factor, type);
	} else {
	return new G1RegionsSmallerThanCommitSizeMapper(rs, actual_size, page_size, region_granularity, commit_factor, type);
	}
	}