1. GARBAGE COLLECTION IN AN UNCOOPERATIVE ENVIRONMENT Hans-Juergen Boehm Computer Science Dept. Rice University, Houston Mark Wieser Xerox Corporation, Palo Alto Presented by Srilakshmi Swati Pendyala

2. Outline  Introduction  Garbage Collection In Different Languages  Problem Domain  Need for conservative garbage collection in uncooperative environments  Overview of the proposed Garbage Collector  Use of the proposed GC as a debugging tool  Implementation Results  Conclusion

3. Introduction  Garbage Collection Different Languages ?  JAVA http://www.folgmann.com/en/gc.html

4. Introduction  Garbage Collection Different Languages : .NET, VB, C# ?  Perl, Python ?  C, C++ ?

5. Introduction  Garbage Collection Different Languages ? .NET, VB, C# – Mark and Sweep, Generational  Perl, Python – Reference Counting  C, C++ – No garbage collection, managed options available.  ADA, Modula 3 – Manual & Automated Garbage Collection

6. Introduction  JAVA,.NET etc.  Automatic Garbage Collection  No memory management effort for the programmer  In the run-time, the program should tell the GC which memory objects are still in use  C, C++ etc.  Program should “free” the allocated memory  Prone to memory leaks etc.  Both cases lead to additional effort from program/compiler.  GC affects the performance of the program  Better performance can be achieved (in some cases) when the program doesn’t worry about GC at all.

7. What is the need to avoid cooperation?  Programmers don’t want to pay for GC unless needed  Disadvantage in tagging the integers  Reduction available number of bits  Difficulty in manipulating standard machine representation of data.  Need for interfacing routines  To implement specific programming language like Russell  To enable garbage collection in conjunction with C, Pascal etc.  Difficult to design compilers that always preserve garbage collection invariants Need for a Garbage Collector that expects less from the program/compiler

8. Uncooperative Environment  Program/compiler does not provide information to recognize pointers  Every register/word potential pointer  All the storage that is accessible by the stack, registers etc., may not be needed by the program  Compiled code may fail to destroy the references (for performance issues/because of bugs)  Particular run-time representations may involve unnecessary references not intended by the programmer  Difficult to tell if an object is actually required by the program  Can lead to program failure if necessary objects are deleted  Need for CONSERVATIVE Garbage Collection

9.  Imagine doing a mark and sweep GC, but not knowing for sure if a cell has a pointer in it or some other data.  If it looks like a pointer (that is, is a valid word- aligned address within heap memory bounds), assume that it IS a pointer, and trace that and other pointers in that record too.  Any heap data that is not marked in this way is garbage and can be collected. (There are no pointers to it.) Conservative Garbage Collection

10. Discussion  Is conservative Garbage Collection needed in Cooperative systems ?  Disadvantages of Conservative Garbage Collection ?  Some amount of inaccessible memory is not reclaimed.  How can we reduce memory lost because of Conservative Garbage Collection ?  Better checks to detect false pointers

11. How does the Garbage Collector work?  Uses Mark-Sweep Stop-the-World Garbage Collection Algorithms  Procedure:  Scan all objects referenced directly by pointer variables (roots) from stack & registers  Verify that pointers are actually pointing to intended objects (validity check) and mark the objects referenced by validated pointers  Mark objects directly reachable from newly marked objects.  Finally identify unmarked objects and free them (sweep) E.g. put them in free lists. Reuse to satisfy allocation requests.  Objects are not moved.

12. Mark/Sweep illustration Stack w/ pointer variables

13. Mark/Sweep illustration (2) Stack w/ pointer variables

14. Allocator design  Allocation scheme obtains “chunks” of memory.  Chunks are always multiples of 4k in size.  Separate free lists for each object size.  Characteristics:  No per object space overhead (except mark bits)  Partial sweeps are possible.

15. Heap layout Free lists...... Heap Data 4k size chunks

16. Data Structure for Chunks  A list of allocated chunks contains pointers to the beginning of each chunk  Contents of a chunk C:  Size of objects in the chunk  A pointer to the entry for C in list of allocated chunks  An area reserved for mark bits corresponding to the objects in the chunk  Data Objects Is it better than “tagging” integers ?

17. Finding Roots & Pointers  Possible roots: registers, stack, static areas  No cooperation from compiler  treat every word as potential pointer  ignore interior pointers (standard)  prefer marking from false pointers over ignoring valid pointers Conservative Pointer Identification: given word p;  does p refer to the collected heap?  does it point into heap block allocated by collector?  does it point to the beginning of an object in that block? if yes,  mark object in block header  push object onto mark stack  Sweep:  If a chunk is completely empty, return it to the chunk allocator

18. Pointer Validity Check Goal: To minimize the marking of false pointers  The pointer “p” should reference to a proper heap- address range for it to correspond to an object  If it corresponds an object, the pointer contained in the chunk header should correspond to the actual address of pointer “p” in the list of allocated chunks  The offset of the supposed object from the chunk header should be a multiple of of the object size given by chunk header and it should be within the end of the chunk

19. Pointer validity check contd.,  Pointer to the interior of an object is viewed as invalid.  Techniques to minimize the probability of accidentally generating an integer that happens to be valid heap address.  Data distinguished as 1) Atomic 2) Composite Atomic : Data that never contains embedded references. Composite : Data that may contain embedded references.

20. Garbage Collector as a Debugging Tool  Use GC to identify allocated memory that is no longer needed by the program, but not yet freed by it.  Use a tracer to track the memory leaks back to the subroutine responsible for them.  Procedure:  An allocation-and-free tracer.  Subroutine names are recorded on a stack with every call to “malloc”.  Mark the storage as freed when ‘free’ calls are made.  When collector runs, storage having no pointers to it and that was never explicitly deallocated with ‘free’ call is likely for storage leak.  Collector running with the tracer could find most of the storage unmarked by the collector, but never been explicitly “free”d.

21. Experimental Results  Mark phase of Russell collector took 1.9 seconds per megabyte of accessible memory in the heap. Sweep phase took 0.4 seconds per megabyte.  Garbage Collection was added to TimberWolf and SDI.  The systems were re-linked so that calls to Unix allocation routines instead called the allocator.  SunView presented problems because of dynamic allocated memory remapping and ‘notifier’.  Programming styles involving disguised pointers will not work with the collector method.  Use of the proposed GC as debugging tool has also been demonstrated on SunView system.

22. Conclusions  GC effective for traditional imperative languages with minimum cooperation from program/compiler  Realistic alternative to explicit memory management for most applications  May not suitable for real-time applications  No big constraints to coding style, except hidden pointer problem  GC’ing allocators competitive even with code not written for GC  The same GC can be used as debugging tool for programs that do manual garbage collection  An implementation of this garbage collector can be downloaded online