Abstract: Concurrent checkpointing for rollback recovery for system failures is disclosed. The system includes a stable database, and a processor configured to receive and process a checkpoint request while a first thread performs a process and a second thread stores contents of memory regions listed in a first list to the stable storage. Processing the checkpoint request includes write protecting all memory regions listed in a previously initialized and populated second list, initializing an empty third list, creating a coalesced list by combining the contents of the first and second lists, and assigning the coalesced list to the second thread while the first thread proceeds with the process.

Abstract: A method of identifying the source of a memory corruption error during operation of a checkpoint library includes receiving an error detection request and, in response to the request, write protecting all memory regions allocated to a checkpoint library. The method further includes detecting when a memory region is accessed for modification during operation of the checkpoint library and, in response to the detection, identifying the source of a memory corruption error affecting the memory region.

Abstract: Concurrent checkpointing for rollback recovery for system failures is disclosed. The system includes a stable database, and a processor configured to receive and process a checkpoint request while a first thread performs a process and a second thread stores contents of memory regions listed in a first list to the stable storage. Processing the checkpoint request includes write protecting all memory regions listed in a previously initialized and populated second list, initializing an empty third list, creating a coalesced list by combining the contents of the first and second lists, and assigning the coalesced list to the second thread while the first thread proceeds with the process.