Modifier and Type	Class and Description
pack-priv class	PooledBufferAllocator.PoolThreadLocalCache

Field Summary

Modifier and Type	Field and Description
private final AllocationType	allocationType
private final List<PoolArenaMetric>	arenaMetrics
private final List<PoolArenaMetric>	arenaMetricsView
private final PoolArena[]	arenas
private final int	chunkSize
private volatile boolean	closed
private static final int	DEFAULT_CACHE_TRIM_INTERVAL
private static final long	DEFAULT_CACHE_TRIM_INTERVAL_MILLIS
private static final int	DEFAULT_DIRECT_MEMORY_CACHE_ALIGNMENT
pack-priv static final int	DEFAULT_MAX_CACHED_BUFFER_CAPACITY
pack-priv static final int	DEFAULT_MAX_CACHED_BYTEBUFFERS_PER_CHUNK
private static final int	DEFAULT_MAX_ORDER
private static final int	DEFAULT_NORMAL_CACHE_SIZE
private static final int	DEFAULT_NUM_DIRECT_ARENA
private static final int	DEFAULT_NUM_HEAP_ARENA
private static final int	DEFAULT_PAGE_SIZE
private static final int	DEFAULT_SMALL_CACHE_SIZE
private static final boolean	DEFAULT_USE_CACHE_FOR_ALL_THREADS
private static final InternalLogger	logger
private final MemoryManager	manager
private static final int	MAX_CHUNK_SIZE
private final PooledBufferAllocatorMetric	metric
private static final int	MIN_PAGE_SIZE
private final int	normalCacheSize
private final AllocatorControl	pooledAllocatorControl
private final int	smallCacheSize
private final PooledBufferAllocator.PoolThreadLocalCache	threadCache
private final Runnable	trimTask

Constructor Summary

Access	Constructor and Description
public	PooledBufferAllocator(MemoryManager manager, boolean direct)
public	PooledBufferAllocator(MemoryManager manager, boolean direct, int numArenas, int pageSize, int maxOrder)
public	PooledBufferAllocator(MemoryManager manager, boolean direct, int numArenas, int pageSize, int maxOrder, int smallCacheSize, int normalCacheSize, boolean useCacheForAllThreads)
public	PooledBufferAllocator(MemoryManager manager, boolean direct, int numArenas, int pageSize, int maxOrder, int smallCacheSize, int normalCacheSize, boolean useCacheForAllThreads, int directMemoryCacheAlignment)

Method Summary

Modifier and Type	Method and Description
public Buffer	allocate(int The size of `Buffer` to allocate. size) Implements io.netty5.buffer.BufferAllocator.allocate. Allocate a `Buffer` of the given size in bytes.
private UntetheredMemory	allocateUnpooled(int size)
pack-priv UntetheredMemory	allocateUntethered(int size)
pack-priv List<PoolArenaMetric>	arenaMetrics() Return a `List` of all heap `PoolArenaMetric`s that are provided by this pool.
pack-priv final int	chunkSize() Return the chunk size for an arena.
public void	close() Implements io.netty5.buffer.BufferAllocator.close. Close this allocator, freeing all of its internal resources.
public Supplier<Buffer>	constBufferSupplier(byte[] The byte contents of the buffers produced by the returned supplier. bytes) Implements io.netty5.buffer.BufferAllocator.constBufferSupplier. Create a supplier of "constant" Buffers from this allocator, that all have the given byte contents.
public static int	defaultMaxOrder() Default maximum order - System Property: io.netty5.allocator.maxOrder - default 11
public static int	defaultNormalCacheSize() Default normal cache size - System Property: io.netty5.allocator.normalCacheSize - default 64
public static int	defaultNumDirectArena() Default number of direct arenas - System Property: io.netty5.allocator.numDirectArenas - default 2 * cores
public static int	defaultNumHeapArena() Default number of heap arenas - System Property: io.netty5.allocator.numHeapArenas - default 2 * cores
public static int	defaultPageSize() Default buffer page size - System Property: io.netty5.allocator.pageSize - default 8192
public static boolean	defaultPreferDirect() Default prefer direct - System Property: io.netty5.noPreferDirect - default false
public static int	defaultSmallCacheSize() Default small cache size - System Property: io.netty5.allocator.smallCacheSize - default 256
public static boolean	defaultUseCacheForAllThreads() Default thread caching behavior - System Property: io.netty5.allocator.useCacheForAllThreads - default true
public String	dumpStats() Returns the status of the allocator (which contains all metrics) as string.
public AllocationType	getAllocationType() Implements io.netty5.buffer.BufferAllocator.getAllocationType. Get the `AllocationType` from this allocator.
pack-priv final AllocatorControl	getPooledAllocatorControl()
public boolean	isDirectBufferPooled()
public static boolean	isDirectMemoryCacheAlignmentSupported() Return `true` if direct memory cache alignment is supported, `false` otherwise.
public boolean	isPooling() Implements io.netty5.buffer.BufferAllocator.isPooling. Determine if this allocator is pooling and reusing its allocated memory.
pack-priv static PoolArena	leastUsedArena(PoolArena[] arenas)
public BufferAllocatorMetric	metric() Implements io.netty5.buffer.pool.BufferAllocatorMetricProvider.metric. Returns a `BufferAllocatorMetric` for a `BufferAllocator`.
private static PoolArena[]	newArenaArray(int size)
pack-priv int	normalCacheSize() Return the size of the normal cache.
public int	numArenas()
pack-priv int	numThreadLocalCaches() Return the number of thread local caches used by this `PooledBufferAllocator`.
pack-priv final long	pinnedMemory()
private static long	pinnedMemory(PoolArena[] arenas)
pack-priv int	smallCacheSize() Return the size of the small cache.
pack-priv final PoolThreadCache	threadCache()
public boolean	Returns: `true` if a cache for the current `Thread` exists and so was trimmed, false otherwise. trimCurrentThreadCache() Trim thread local cache for the current `Thread`, which will give back any cached memory that was not allocated frequently since the last trim operation.
pack-priv final long	usedMemory()
private static long	usedMemory(PoolArena[] arenas)
private static int	validateAndCalculateChunkSize(int pageSize, int maxOrder)
private static int	validateAndCalculatePageShifts(int pageSize, int alignment)

Inherited from java.lang.Object:: clone equals finalize getClass hashCode notify notifyAll toString wait wait wait

Field Detail

allocationType	back to summary
private final AllocationType allocationType

arenaMetrics	back to summary
private final List<PoolArenaMetric> arenaMetrics

arenaMetricsView	back to summary
private final List<PoolArenaMetric> arenaMetricsView

arenas	back to summary
private final PoolArena[] arenas

chunkSize	back to summary
private final int chunkSize

closed	back to summary
private volatile boolean closed

DEFAULT_CACHE_TRIM_INTERVAL	back to summary
private static final int DEFAULT_CACHE_TRIM_INTERVAL

DEFAULT_CACHE_TRIM_INTERVAL_MILLIS	back to summary
private static final long DEFAULT_CACHE_TRIM_INTERVAL_MILLIS

DEFAULT_DIRECT_MEMORY_CACHE_ALIGNMENT	back to summary
private static final int DEFAULT_DIRECT_MEMORY_CACHE_ALIGNMENT

DEFAULT_MAX_CACHED_BUFFER_CAPACITY	back to summary
pack-priv static final int DEFAULT_MAX_CACHED_BUFFER_CAPACITY

DEFAULT_MAX_CACHED_BYTEBUFFERS_PER_CHUNK	back to summary
pack-priv static final int DEFAULT_MAX_CACHED_BYTEBUFFERS_PER_CHUNK

DEFAULT_MAX_ORDER	back to summary
private static final int DEFAULT_MAX_ORDER

DEFAULT_NORMAL_CACHE_SIZE	back to summary
private static final int DEFAULT_NORMAL_CACHE_SIZE

DEFAULT_NUM_DIRECT_ARENA	back to summary
private static final int DEFAULT_NUM_DIRECT_ARENA

DEFAULT_NUM_HEAP_ARENA	back to summary
private static final int DEFAULT_NUM_HEAP_ARENA

DEFAULT_PAGE_SIZE	back to summary
private static final int DEFAULT_PAGE_SIZE

DEFAULT_SMALL_CACHE_SIZE	back to summary
private static final int DEFAULT_SMALL_CACHE_SIZE

DEFAULT_USE_CACHE_FOR_ALL_THREADS	back to summary
private static final boolean DEFAULT_USE_CACHE_FOR_ALL_THREADS

logger	back to summary
private static final InternalLogger logger

manager	back to summary
private final MemoryManager manager

MAX_CHUNK_SIZE	back to summary
private static final int MAX_CHUNK_SIZE

metric	back to summary
private final PooledBufferAllocatorMetric metric

MIN_PAGE_SIZE	back to summary
private static final int MIN_PAGE_SIZE

normalCacheSize	back to summary
private final int normalCacheSize

pooledAllocatorControl	back to summary
private final AllocatorControl pooledAllocatorControl

smallCacheSize	back to summary
private final int smallCacheSize

threadCache	back to summary
private final PooledBufferAllocator.PoolThreadLocalCache threadCache

trimTask	back to summary
private final Runnable trimTask

Constructor Detail

PooledBufferAllocator	back to summary
public PooledBufferAllocator(MemoryManager manager, boolean direct)

PooledBufferAllocator	back to summary
public PooledBufferAllocator(MemoryManager manager, boolean direct, int numArenas, int pageSize, int maxOrder)

PooledBufferAllocator	back to summary
public PooledBufferAllocator(MemoryManager manager, boolean direct, int numArenas, int pageSize, int maxOrder, int smallCacheSize, int normalCacheSize, boolean useCacheForAllThreads)

PooledBufferAllocator	back to summary
public PooledBufferAllocator(MemoryManager manager, boolean direct, int numArenas, int pageSize, int maxOrder, int smallCacheSize, int normalCacheSize, boolean useCacheForAllThreads, int directMemoryCacheAlignment)

Method Detail

allocate back to summary

allocate	back to summary
public Buffer allocate(int size) Implements io.netty5.buffer.BufferAllocator.allocate. Doc from io.netty5.buffer.BufferAllocator.allocate. Allocate a `Buffer` of the given size in bytes. This method may throw an `OutOfMemoryError` if there is not enough free memory available to allocate a `Buffer` of the requested size. The buffer will use big endian byte order. Note unlike the JDK `ByteBuffer`s, Netty `Buffers` are not guaranteed to be zeroed when allocated. In other words, the memory of a newly allocated buffer may contain garbage data from prior allocations, and the memory is likewise not guaranteed to be erased when the buffer is closed. If the data is sensitive and needs to be overwritten when the buffer is closed, then the buffer should be allocated with the `SensitiveBufferAllocator`. Parameters size:int The size of `Buffer` to allocate. Returns:Buffer The newly allocated `Buffer`. Annotations @Override

public Buffer allocate(int size)

Implements io.netty5.buffer.BufferAllocator.allocate.

Doc from io.netty5.buffer.BufferAllocator.allocate.

Allocate a Buffer of the given size in bytes. This method may throw an OutOfMemoryError if there is not enough free memory available to allocate a Buffer of the requested size.

The buffer will use big endian byte order.

Note

unlike the JDK ByteBuffers, Netty Buffers are not guaranteed to be zeroed when allocated. In other words, the memory of a newly allocated buffer may contain garbage data from prior allocations, and the memory is likewise not guaranteed to be erased when the buffer is closed. If the data is sensitive and needs to be overwritten when the buffer is closed, then the buffer should be allocated with the SensitiveBufferAllocator.

Parameters

size:int: The size of Buffer to allocate.

Returns:Buffer

The newly allocated Buffer.

Annotations

@Override

allocateUnpooled	back to summary
private UntetheredMemory allocateUnpooled(int size)

allocateUntethered	back to summary
pack-priv UntetheredMemory allocateUntethered(int size)

arenaMetrics	back to summary
pack-priv List<PoolArenaMetric> arenaMetrics() Return a `List` of all heap `PoolArenaMetric`s that are provided by this pool.

chunkSize	back to summary
pack-priv final int chunkSize() Return the chunk size for an arena.

close back to summary

close	back to summary
public void close() Implements io.netty5.buffer.BufferAllocator.close. Doc from io.netty5.buffer.BufferAllocator.close. Close this allocator, freeing all of its internal resources. Existing (currently in-use) allocated buffers will not be impacted by calling this method. If this is a pooling or caching allocator, then existing buffers will be immediately freed when they are closed, instead of being pooled or cached. The allocator can no longer be used to allocate more buffers after calling this method. Attempting to allocate from a closed allocator will cause `IllegalStateException`s to be thrown. Annotations @Override

public void close()

Implements io.netty5.buffer.BufferAllocator.close.

Doc from io.netty5.buffer.BufferAllocator.close.

Close this allocator, freeing all of its internal resources.

Existing (currently in-use) allocated buffers will not be impacted by calling this method. If this is a pooling or caching allocator, then existing buffers will be immediately freed when they are closed, instead of being pooled or cached.

The allocator can no longer be used to allocate more buffers after calling this method. Attempting to allocate from a closed allocator will cause IllegalStateExceptions to be thrown.

Annotations

@Override

constBufferSupplier back to summary

constBufferSupplier	back to summary
public Supplier<Buffer> constBufferSupplier(byte[] bytes) Implements io.netty5.buffer.BufferAllocator.constBufferSupplier. Doc from io.netty5.buffer.BufferAllocator.constBufferSupplier. Create a supplier of "constant" Buffers from this allocator, that all have the given byte contents. The buffer has the same capacity as the byte array length, and its write offset is placed at the end, and its read offset is at the beginning, such that the entire buffer contents are readable. The buffers produced by the supplier will each have their own independent life-cycle, and closing them will make them inaccessible, just like normally allocated buffers. The buffers produced are "constants", in the sense that they are read-only. It can generally be expected, but is not guaranteed, that the returned supplier is more resource efficient than allocating and copying memory with other available APIs. In such optimised implementations, the underlying memory baking the buffers will be shared among all the buffers produced by the supplier. The primary use case for this API, is when you need to repeatedly produce buffers with the same contents, and you perhaps wish to keep a `static final` field with these contents. The supplier-based API enforces that each usage get their own distinct buffer instance. Each of these instances cannot interfere with each other, so bugs like closing, or modifying the contents, of a shared buffer cannot occur. Parameters bytes:byte[] The byte contents of the buffers produced by the returned supplier. Returns:Supplier<Buffer> A supplier of read-only buffers with the given contents. Annotations @Override

public Supplier<Buffer> constBufferSupplier(byte[] bytes)

Implements io.netty5.buffer.BufferAllocator.constBufferSupplier.

Doc from io.netty5.buffer.BufferAllocator.constBufferSupplier.

Create a supplier of "constant" Buffers from this allocator, that all have the given byte contents. The buffer has the same capacity as the byte array length, and its write offset is placed at the end, and its read offset is at the beginning, such that the entire buffer contents are readable.

The buffers produced by the supplier will each have their own independent life-cycle, and closing them will make them inaccessible, just like normally allocated buffers.

The buffers produced are "constants", in the sense that they are read-only.

It can generally be expected, but is not guaranteed, that the returned supplier is more resource efficient than allocating and copying memory with other available APIs. In such optimised implementations, the underlying memory baking the buffers will be shared among all the buffers produced by the supplier.

The primary use case for this API, is when you need to repeatedly produce buffers with the same contents, and you perhaps wish to keep a static final field with these contents. The supplier-based API enforces that each usage get their own distinct buffer instance. Each of these instances cannot interfere with each other, so bugs like closing, or modifying the contents, of a shared buffer cannot occur.

Parameters

bytes:byte[]: The byte contents of the buffers produced by the returned supplier.

Returns:Supplier<Buffer>

A supplier of read-only buffers with the given contents.

Annotations

@Override

defaultMaxOrder	back to summary
public static int defaultMaxOrder() Default maximum order - System Property: io.netty5.allocator.maxOrder - default 11

defaultNormalCacheSize	back to summary
public static int defaultNormalCacheSize() Default normal cache size - System Property: io.netty5.allocator.normalCacheSize - default 64

defaultNumDirectArena	back to summary
public static int defaultNumDirectArena() Default number of direct arenas - System Property: io.netty5.allocator.numDirectArenas - default 2 * cores

defaultNumHeapArena	back to summary
public static int defaultNumHeapArena() Default number of heap arenas - System Property: io.netty5.allocator.numHeapArenas - default 2 * cores

defaultPageSize	back to summary
public static int defaultPageSize() Default buffer page size - System Property: io.netty5.allocator.pageSize - default 8192

defaultPreferDirect	back to summary
public static boolean defaultPreferDirect() Default prefer direct - System Property: io.netty5.noPreferDirect - default false

defaultSmallCacheSize	back to summary
public static int defaultSmallCacheSize() Default small cache size - System Property: io.netty5.allocator.smallCacheSize - default 256

defaultUseCacheForAllThreads	back to summary
public static boolean defaultUseCacheForAllThreads() Default thread caching behavior - System Property: io.netty5.allocator.useCacheForAllThreads - default true

dumpStats	back to summary
public String dumpStats() Returns the status of the allocator (which contains all metrics) as string. Be aware this may be expensive and so should not be called too frequently.

getAllocationType back to summary

public AllocationType getAllocationType()

Implements io.netty5.buffer.BufferAllocator.getAllocationType.

Doc from io.netty5.buffer.BufferAllocator.getAllocationType.

Get the AllocationType from this allocator. This would typically be one of the StandardAllocationTypes.

Returns:AllocationType

The type of allocations performed by this allocator.

Annotations

@Override

getPooledAllocatorControl	back to summary
pack-priv final AllocatorControl getPooledAllocatorControl()

isDirectBufferPooled	back to summary
public boolean isDirectBufferPooled()

isDirectMemoryCacheAlignmentSupported	back to summary
public static boolean isDirectMemoryCacheAlignmentSupported() Return `true` if direct memory cache alignment is supported, `false` otherwise.

isPooling back to summary

public boolean isPooling()

Implements io.netty5.buffer.BufferAllocator.isPooling.

Doc from io.netty5.buffer.BufferAllocator.isPooling.

Determine if this allocator is pooling and reusing its allocated memory.

Returns:boolean

true if this allocator is pooling and reusing its memory, false otherwise.

Annotations

@Override

leastUsedArena	back to summary
pack-priv static PoolArena leastUsedArena(PoolArena[] arenas)

metric back to summary

public BufferAllocatorMetric metric()

Implements io.netty5.buffer.pool.BufferAllocatorMetricProvider.metric.

Doc from io.netty5.buffer.pool.BufferAllocatorMetricProvider.metric.

Returns a BufferAllocatorMetric for a BufferAllocator.

Annotations

@Override

newArenaArray	back to summary
private static PoolArena[] newArenaArray(int size)

normalCacheSize	back to summary
pack-priv int normalCacheSize() Return the size of the normal cache.

numArenas	back to summary
public int numArenas()

numThreadLocalCaches	back to summary
pack-priv int numThreadLocalCaches() Return the number of thread local caches used by this `PooledBufferAllocator`.

pinnedMemory	back to summary
pack-priv final long pinnedMemory()

pinnedMemory	back to summary
private static long pinnedMemory(PoolArena[] arenas)

smallCacheSize	back to summary
pack-priv int smallCacheSize() Return the size of the small cache.

threadCache	back to summary
pack-priv final PoolThreadCache threadCache()

trimCurrentThreadCache back to summary

public boolean trimCurrentThreadCache()

Trim thread local cache for the current Thread, which will give back any cached memory that was not allocated frequently since the last trim operation.

Returns:boolean: true if a cache for the current Thread exists and so was trimmed, false otherwise.

usedMemory	back to summary
pack-priv final long usedMemory()

usedMemory	back to summary
private static long usedMemory(PoolArena[] arenas)

validateAndCalculateChunkSize	back to summary
private static int validateAndCalculateChunkSize(int pageSize, int maxOrder)

validateAndCalculatePageShifts	back to summary
private static int validateAndCalculatePageShifts(int pageSize, int alignment)

io.netty5.buffer.pool back to summary

pack-priv final Class PooledBufferAllocator.PoolThreadLocalCache

Class Inheritance

extends FastThreadLocal<PoolThreadCache>

java.lang.Object
io.netty5.util.concurrent.FastThreadLocal
io.netty5.buffer.pool.PooledBufferAllocator.PoolThreadLocalCache

Field Summary

Modifier and Type	Field and Description
private final boolean	useCacheForAllThreads

Constructor Summary

Access	Constructor and Description
pack-priv	PoolThreadLocalCache(boolean useCacheForAllThreads)

Method Summary

Modifier and Type	Method and Description
protected synchronized PoolThreadCache	initialValue() Overrides io.netty5.util.concurrent.FastThreadLocal.initialValue. Returns the initial value for this thread-local variable.
protected void	onRemoval(PoolThreadCache threadCache) Overrides io.netty5.util.concurrent.FastThreadLocal.onRemoval. Invoked when this thread local variable is removed by `remove()`.

Inherited from io.netty5.util.concurrent.FastThreadLocal:: destroy get getIfExists isSet remove removeAll set size

Field Detail

useCacheForAllThreads	back to summary
private final boolean useCacheForAllThreads

Constructor Detail

PoolThreadLocalCache	back to summary
pack-priv PoolThreadLocalCache(boolean useCacheForAllThreads)

Method Detail

initialValue

back to summary

protected synchronized PoolThreadCache initialValue()

Overrides io.netty5.util.concurrent.FastThreadLocal.initialValue.

Doc from io.netty5.util.concurrent.FastThreadLocal.initialValue.

Returns the initial value for this thread-local variable.

Annotations

@Override

onRemoval back to summary

protected void onRemoval(PoolThreadCache threadCache)

Overrides io.netty5.util.concurrent.FastThreadLocal.onRemoval.

Doc from io.netty5.util.concurrent.FastThreadLocal.onRemoval.

Invoked when this thread local variable is removed by remove(). Be aware that remove() is not guaranteed to be called when the `Thread` completes which means you can not depend on this for cleanup of the resources in the case of `Thread` completion.

Annotations

@Override

public Class PooledBufferAllocator

Nested and Inner Type Summary

Field Summary

Constructor Summary

Method Summary

Field Detail

Constructor Detail

Method Detail

pack-priv final Class PooledBufferAllocator.PoolThreadLocalCache

Field Summary

Constructor Summary

Method Summary

Field Detail

Constructor Detail

Method Detail