avail/avail.interpreter.levelTwo/L2Chunk/Companion

Companion

Functions

fun allocate( code: A_RawFunction?, numObjects: Int, numIntegers: Int, numFloats: Int, offsetAfterInitialTryPrimitive: Int, theInstructions: List<L2Instruction>, controlFlowGraph: L2ControlFlowGraph, contingentValues: A_Set): L2Chunk

Allocate and set up a new L2Chunk with the given information. If code is non-null, set it up to use the new chunk for subsequent invocations.

Properties

countdownForNewCode

const val countdownForNewCode: Long = 10000

Return the number of times to invoke a compiled code object, after creation, before attempting to optimize it for the first time.

countdownForNewlyOptimizedCode

const val countdownForNewlyOptimizedCode: Long

Return the number of times to invoke a compiled code object, after optimization, before attempting to optimize it again with more effort.

decrementForPolledActiveCode

const val decrementForPolledActiveCode: Long = 1000

Each time an A_RawFunction is found to be the running code for some interpreter during periodic polling, atomically decrease its countdown by this amount, avoiding going below one (1).

invalidationLock

val invalidationLock: ReentrantLock

The lock that protects invalidation of chunks due to method changes from interfering with each other. The alternative to a global lock seems to imply deadlock conditions.

unoptimizedChunk

@JvmField

val unoptimizedChunk: L2Chunk

The special level two chunk that is used to interpret level one nybblecodes until a piece of compiled code has been executed some number of times (specified in countdownForNewCode).