International Business Machines Corporation

Ryan Andrew Sciampacone

Aleksandar Micic

Peter Wiebe Burka

Jeffrey Michael Disher

Daryl James Maier

Vijay Sundaresan

Curtis E Hrischuk

Angela Lin

Trent Gray-Donald

Karl Michael Taylor

Peter Duncan Shipton

Berthold Martin Lebert

Elijah El-Haddad

John Duimovich

Kim Clohessy

David Kevin Siegwart

Michael S Fulton

Charles Robert Gracie

Nikola Grcevski

Gary John DeVal

Henry W Adams, Iii

Aleksander Micic

A technique for incrementally unloading classes using a region-based garbage collector is described. One aspect of such a technique includes maintaining a remembered set for a class set. The remembered set indicates whether instances of the class set are contained in one or more regions in memory, and in which regions the instances are contained. Upon performing an incremental garbage collection process for a subset of the regions in memory, the technique examines the remembered set to determine whether the class set includes instances in regions outside of the subset. If the remembered set indicates that the class set includes instances outside of the subset of regions, the technique identifies the class set as 'live.' This will preclude unloading the class set from the subset of regions.

Incremental class unloading in a region-based garbage collector

A method for creating an enhanced array is provided. An enhanced compiler generates an instrumented class file for an enhanced array class. The enhanced array class comprises at least one element class. The enhanced runtime loads the instrumented class file of the enhanced array class. The enhanced runtime loads a class file corresponding to the element class of the enhanced array class, and builds an internal data structure corresponding to the enhanced array class. An amount of memory storage is calculated for storing the enhanced array instance, and the amount of memory storage is allocated.

Enhanced array class

Enhanced java array

An illustrative embodiment of a computer-implemented process for managing a staging area creates the staging area for identified candidate cold objects, moves the identified candidate objects into the staging area, tracks application access to memory comprising the staging area and determines whether frequency of use information for a specific object exceeds a predetermined threshold. Responsive to a determination that the frequency of use information for the specific object exceeds a predetermined threshold, move the specific object into a regular area and determine whether a current time exceeds a predetermined threshold. Responsive to a determination that the current time exceeds a predetermined threshold, the computer-implemented process moves remaining objects from the staging area to a cold area.

Data staging area

A computer-implemented process for write barrier elision during program execution receives a request to overwrite a reference, determines whether garbage collection is active and responsive to a determination that garbage collection is active, determines whether a reference object has been scanned. Responsive to a determination that the reference object has not been scanned set a mark bit for the reference object in a modified mark map, the reference object is scanned and a scan bit is set for the reference object in the modified mark map.

Write barrier elision for reference arrays

At least one region of a heap that includes memory allocations is analyzed, using age and occupancy criteria, across a number of local garbage collection cycles using a processor executing a region-based garbage collector. Based upon the analyzed age and occupancy criteria of the at least one region, at least one stable region in age and occupancy is identified among the at least one region of the heap across the number of local garbage collection cycles. Maintenance of a remembered set (RS) of external references into the at least one stable region is temporarily stopped for each identified stable region during at least one additional local garbage collection cycle.

Remembered set overhead reduction by deferred garbage collection of stable regions

A method to implement parallel, single-pass compaction in a garbage collector is described. In one embodiment, such a method includes conducting a planning phase for multiple regions to be compacted. During the planning phase, the method determines new locations for data entities in the multiple regions. The method then performs a move phase for the multiple regions to move the data entities to their new locations. During the move phase, the method initiates multiple compaction threads to move the data entities to their new locations. While executing, the compaction threads dynamically build a dependency graph of the regions being compacted. The dependency graph guarantees that no data entity is moved to its new location until all data entities that it overwrites have been moved to their new locations. A corresponding computer program product and apparatus are also disclosed herein.

Parallel, single-pass compaction in a region-based garbage collector

Automatic memory management with a generational copy collector garbage collection algorithm running on parallel threads to perform simultaneous and independent copying and scanning. An object is scanned. It is determined whether an object referenced within the slot of the object being scanned may be copied to the same generation as the scanned object. A mark may be made to indicate that the referenced object may be copied to a different generation than the scanned object. The mark defers the copying of the referenced object.

Defer separating children in parallel copying garbage collection

Remembered set overhead reduction by deferred garbage collections of stable regions

An illustrative embodiment of a computer-implemented process for compaction planning selects a source region from a set of regions to form a selected source region, initializes process data using information from the selected source region and responsive to a determination that a current destination is not NULL and not full, determines whether to atomically consume from a destination. Responsive to a determination to atomically consume from a destination, the computer-implemented process plans to evacuate into a consumed extent and updates the process data and responsive to a determination that the source region is empty, determines whether more work remains. Responsive to a determination that more work does not remain, the computer-implemented process generates a relocation table.

Compaction planning

Tracking newly created objects during a garbage collection cycle includes marking newly allocated objects in the GC cycle during the allocation cache population phase. The goal is to take a high performance system for tracing live objects, such as the mark map, and combine the SATB with a cached allocation system for heap memory such that SATB representation is preserved. Embodiments combine both efficient support for an SATB model during a collection cycle and a highly optimized allocation path for heap objects in an allocation scheme (which may optionally be a size segregated class allocation scheme) that allows both mechanisms to co-exist effectively. Specifically, embodiments track newly created objects during a GC cycle such that they are kept alive during the trace phase while continuing to support a highly efficient allocating caching strategy.

Allocation cache premarking for snap-shot-at-the-beginning concurrent mark-and-sweep collector

An illustrative embodiment of a computer-implemented process for dynamic measurement and adaptation of a parallel copying garbage collector initializes values and data structures, receives an evacuate set, determines whether a new copy block is required and responsive to a determination that a new copy block is required, calculates a size of the new copy block. The computer-implemented process further selects a survivor region from a list of survivor regions to form a selected survivor region, places a lock on the selected survivor region, identifies a number of threads enqueued on the selected survivor region, responsive to a determination that a contention value exceeds a predetermined value, increases a size of the list of survivor regions, evacuates blocks to the selected survivor region and responsive to a determination that more evacuate blocks do not exist, terminates.

Dynamic measurement and adaptation of a copying garbage collector

An operating system or virtual machine of an information handling system (IHS) initializes a garbage collector to provide object memory compaction during application execution. The operating system or virtual machine performs move and fixup management within heap or object memory stores. The garbage collector identifies holes or unused portions of heap memory and moves one or more objects during compaction operations. After the garbage collector moves the object, a fixup cache tool fixes up the object pointers within the object. The fixup cache tool maintains a fixup cache that includes pointer information corresponding to the objects that the garbage collector moves. The fixup cache tool employs the pointer information within the fixup cache to update or otherwise fixup the objects within the heap that the garbage collector moves during compaction operations. The fixup cache may provide predictive pointer update information during fixup operations for each object move within the heap.

Growing community of inventors

Nepean, Canada

Ryan Andrew Sciampacone