Memory Management for Self-stabilizing Operating Systems

Stabilizing Trust and Reputation for Self-Stabilizing Efficient Hosts in Spite of Byzantine Guests (Extended Abstract)

“…In previous works [19,20,21], we presented new concepts and directions for building a self stabilizing operating system kernel. A self-stabilizing algorithm/system makes the obvious assumption that it is executed.…”

“…Definitions. We briefly define the system states and state transitions (see [19,20] for details concerning processor executions, interrupt, registers, read-only memories, a watchdog and additional settings). A state of the system is an assignment to its various memory components (including the program counter resister).…”

“…(b) A self-stabilizing host operating system [19] which is guaranteed to periodically run some boot-code loaded in a secure way [2,45,51], without being subverted ( [43]) (c) This trusted operating system guarantees eventual execution of all runnable processes including the contract offline detectors. (d) Code is being refreshed ( [19,20]) periodically, so Byzantine or wrong behavior caused by transient faults to code segments are eventually fixed. (e) Contract properties are asserted by online enforcers.…”

“…In such a case when recognizing Byzantine behavior, apart from preventing this behavior and refreshing the code, the closest state (using Hamming distance or some other metric) can be applied to the program. (g) All resource allocations are granted using leases with a self-stabilizing manager, as demonstrated in a previous work on dynamic memory [20], ensuring that resource allocations are eventually fair. The contract detectors and enforcers check also for behavior which violates the leasing rules.…”

“…In previous works we demonstrated the construction of a self-stabilizing operating system (sos) [19,20,21,47]. Guest separation was achieved by using the segmentation mechanism of the Pentium processor [31], without mmu hardware protection.…”