Package io.netty.channel.uring

Class IoUringBufferRing

java.lang.Object

io.netty.channel.uring.IoUringBufferRing

final class IoUringBufferRing
extends java.lang.Object

Nested Class Summary

Nested Classes

Modifier and Type	Class	Description
private class	IoUringBufferRing.RingConsumer

Field Summary

Fields

Modifier and Type	Field	Description
private int	allocatedBuffers
private IoUringBufferRingAllocator	allocator
private boolean	batchAllocation
private int	batchSize
private short	bufferGroupId
private ByteBuf[]	buffers
private boolean	closed
private boolean	corrupted
private short	entries
private IoUringBufferRingExhaustedEvent	exhaustedEvent
private boolean	incremental
private java.nio.ByteBuffer	ioUringBufRing
private short	lastGeneratedBid
private short	mask
private boolean	needExpand
private IoUringBufferRing.RingConsumer	ringConsumer
private int	ringFd
private static VarHandle	SHORT_HANDLE
private int	tailFieldPosition
private int	usableBuffers

Constructor Summary

Constructors

Constructor	Description
IoUringBufferRing(int ringFd, java.nio.ByteBuffer ioUringBufRing, short entries, int batchSize, short bufferGroupId, boolean incremental, IoUringBufferRingAllocator allocator, boolean batchAllocation)

Method Summary

Modifier and Type	Method	Description
(package private) int	allocatedBuffers()	The number of buffers that are currently allocated (and so posted) in this buffer ring.
(package private) int	attemptedBytesRead(short bid)	Return the amount of bytes that we attempted to read for the given id.
(package private) short	bufferGroupId()	The group id that is assigned to this buffer ring.
private int	calculateNextBufferBatch()
(package private) void	close()	Close this IoUringBufferRing, using it after this method is called will lead to undefined behaviour.
(package private) boolean	expand()	Try to expand by adding more buffers to the ring if there is any space left, this will be done lazy.
private void	fill(short bid)
private void	fill(short startBid, int buffers)
(package private) IoUringBufferRingExhaustedEvent	getExhaustedEvent()
(package private) void	initialize()
(package private) boolean	isUsable()
(package private) short	nextBid(short bid, int allocatedBuffers)	Return the next buffer id that follows bid in ring order, wrapping around after allocatedBuffers ids.
(package private) ByteBuf	useBuffer(short bid, int read, boolean more)	Use the buffer for the given buffer id.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

SHORT_HANDLE

private static final VarHandle SHORT_HANDLE

ioUringBufRing

private final java.nio.ByteBuffer ioUringBufRing

tailFieldPosition

private final int tailFieldPosition

entries

private final short entries

mask

private final short mask

bufferGroupId

private final short bufferGroupId

ringFd

private final int ringFd

buffers

private final ByteBuf[] buffers

allocator

private final IoUringBufferRingAllocator allocator

batchAllocation

private final boolean batchAllocation

exhaustedEvent

private final IoUringBufferRingExhaustedEvent exhaustedEvent

ringConsumer

private final IoUringBufferRing.RingConsumer ringConsumer

incremental

private final boolean incremental

batchSize

private final int batchSize

corrupted

private boolean corrupted

closed

private boolean closed

usableBuffers

private int usableBuffers

allocatedBuffers

private int allocatedBuffers

needExpand

private boolean needExpand

lastGeneratedBid

private short lastGeneratedBid

Constructor Detail

IoUringBufferRing

IoUringBufferRing(int ringFd,
                  java.nio.ByteBuffer ioUringBufRing,
                  short entries,
                  int batchSize,
                  short bufferGroupId,
                  boolean incremental,
                  IoUringBufferRingAllocator allocator,
                  boolean batchAllocation)

Method Detail

isUsable

boolean isUsable()

initialize

void initialize()

expand

boolean expand()

Try to expand by adding more buffers to the ring if there is any space left, this will be done lazy.

Returns:

true if we can expand the number of buffers in the ring, false otherwise.

fill

private void fill(short startBid,
                  int buffers)

fill

private void fill(short bid)

getExhaustedEvent

IoUringBufferRingExhaustedEvent getExhaustedEvent()

Returns:

the IoUringBufferRingExhaustedEvent that should be used to signal that there were no buffers left for this buffer ring.

attemptedBytesRead

int attemptedBytesRead(short bid)

Return the amount of bytes that we attempted to read for the given id. This method must be called before useBuffer(short, int, boolean).

Parameters:

bid - the id of the buffer.

Returns:

the attempted bytes.

calculateNextBufferBatch

private int calculateNextBufferBatch()

useBuffer

ByteBuf useBuffer(short bid,
                  int read,
                  boolean more)

Use the buffer for the given buffer id. The returned ByteBuf must be released once not used anymore.

Parameters:

bid - the id of the buffer

read - the number of bytes that could be read. This value might be larger then what a single ByteBuf can hold. Because of this, the caller should call

allocatedBuffers

int allocatedBuffers()

The number of buffers that are currently allocated (and so posted) in this buffer ring. This value can change every time useBuffer(short, int, boolean) is called as it might grow the ring. Because of this, callers that need to walk a sequence of consecutive buffer ids that belong to the same completion (for example a IORING_RECVSEND_BUNDLE completion) must capture this value before calling useBuffer(short, int, boolean) and pass it to nextBid(short, int) afterwards, so the wrap-around is computed using the size of the ring as the kernel observed it when it produced the bundle, not the (possibly already grown) size afterwards.

Returns:

the number of allocated buffers.

nextBid

short nextBid(short bid,
              int allocatedBuffers)

Return the next buffer id that follows bid in ring order, wrapping around after allocatedBuffers ids.

Parameters:

bid - the current buffer id.

allocatedBuffers - the number of allocated buffers to use for the wrap-around computation. This should be a value obtained from allocatedBuffers() before the corresponding useBuffer(short, int, boolean) call for bid, as that call might have grown the ring already.

Returns:

the next buffer id.

bufferGroupId

short bufferGroupId()

The group id that is assigned to this buffer ring.

Returns:

group id.

close

void close()

Close this IoUringBufferRing, using it after this method is called will lead to undefined behaviour.