Major heap Minor heaps Stacks 1 2 3

We describe and prove the correctness of a new concurrent mark-and-sweep garbage collection algorithm. This algorithm derives from the classical on-they algorithm from Dijkstra et al. 9]. A distinguishing feature of our algorithm is that it supports multiprocessor environments where the registers of running processes are not readily accessible, without imposing any overhead on the elementary operations of loading a register or reading or initializing a eld. Furthermore our collector never blocks running mutator processes except possibly on requests for free memory; in particular , updating a eld or creating or marking or sweeping a heap object does not involve system-dependent synchronization primitives such as locks. We also provide support for process creation and deletion, and for managing an ex-tensible heap of variable-sized objects. 1 Introduction Concurrent garbage collection has a well-deserved reputation for being a tough problem. This is evidenced by the discrepancies between the state of theory and practice in this area. As we shall see below, the published proven algorithms often contain simplifying assumptions that cannot be met in practice in a multiprocessor system, because this would either impose unbearable overhead on the mutator processes, or require a degree of hardware and/or operating system support that compromises portability. Implemented systems that do not fall in the latter two categories often rely on incompletely formalized algorithms, which generally means buggy algorithms, given the subtleness of the correct-ness proofs. To our knowledge, and as we shall argue below, all published concurrent collectors fall in one of the above categories , and thus fail to meet at least one of the basic requirements for portable, eeective garbage collection on mul-tiprocessors. In fact the only proposal that even attempts to meet these requirements is the Doligez-Leroy hybrid collector 10]. Unfortunately, the algorithm they proposed was incompletely speciied and, perhaps not unexpectedly, buggy.