Accurate Garbage Collection in Uncooperative Environments with Lazy Pointer Stacks

Abstract

Abstract. Implementing a new programming language by the means of a translator to an existing language is attractive as it provides portability over all platforms supported by the host language and reduces the development time as many low-level tasks can be delegated to the host compiler. The C and C++ programming languages are popular choices for many language implementations due to the availability of efficient compilers on many platforms, and good portability. For garbage-collected languages, however, they are not a perfect match as they provide no support for accurately discovering pointers to heap-allocated data. We evaluate the published techniques, and propose a new mechanism, lazy pointer stacks, for performing accurate garbage collection in such uncooperative environments. We implemented the new technique in the Ovm Java virtual machine with our own Java-to-C++ compiler and GCC as a back-end, and found that our technique outperforms existing approaches.

Precise garbage collection for C

“…In particular, the shadow stack that we use for GC frame is the same as in Henderson's system. Baker et al [2007] demonstrated an improvement to Henderson's linked stack-frame technique by putting stack-variable addresses in a separate array, instead of embedding frames in the C stack, but also by using C++ exceptions to take pointer references only when a collection occurs. Jung et al [2006] use a similar approach; local variables are updated upon return from a function that performed a garbage collection.…”

Garbage collection for safety critical Java

“…For a high-priority GC this point can be at each bytecode (or at each machine instruction for compiling Java). The auxiliary data structure to capture the stack layout (and information which machine register will hold a reference) can get quite large or require additional effort to compute [6].…”

“…From that point on the GC runs in concurrent mode in it's own thread and omits scanning of the thread stacks. 6 We have implemented it for array access, but applying this optimization for field access is straight forward. 7 Since objects are typically small, this optimization is likely to pay off only for arrays.…”