Abstract:
One or more embodiments may provide a method for performing a replay. The method includes initiating execution of a program, the program having a plurality of sets of instructions, and each set of instructions has a number of chunks of instructions. The method also includes intercepting, by a virtual machine unit executing on a processor, an instruction of a chunk of the number of chunks before execution. The method further includes determining, by a replay module executing on the processor, whether the chunk is an active chunk, and responsive to the chunk being the active chunk, executing the instruction.
Abstract:
A serverless distributed file system manages the storage of files and directories using one or more directory groups. The directories may be managed using Byzantine-fault-tolerant groups, whereas files are managed without using Byzantine-fault-tolerant groups. Additionally, the file system may employ a hierarchical namespace to store files. Furthermore, the directory group may employ a plurality of locks to control access to objects (e.g., files and directories) in each directory.
Abstract:
An apparatus and method, using an inter-processor lock to control access to inter-process relationship data structures in the memory (3a, 3b, ..., 3n) of each processor (2a, 2b, ..., 2n) in a multiprocessor system (1). The apparatus and method insure that each inter-process relationship is modified in the same sequence on each processor (2a, 2b, ..., 2n). The apparatus and method also insure that an inter-process relationship is maintained in a consistent state in the face of failure of any of the processors (2a, 2b, ..., 2n).
Abstract:
An apparatus and method, using an inter-processor lock to control access to inter-process relationship data structures in the memory (3a, 3b, ..., 3n) of each processor (2a, 2b, ..., 2n) in a multiprocessor system (1). The apparatus and method insure that each inter-process relationship is modified in the same sequence on each processor (2a, 2b, ..., 2n). The apparatus and method also insure that an inter-process relationship is maintained in a consistent state in the face of failure of any of the processors (2a, 2b, ..., 2n).
Abstract:
Recovering from failure of a distributed processing system process designated as a master process for at least one shared resource. The method and system of the invention provides for detection of the failure (200) by one or more of the shadow processes. The detecting process tests (202) to determine whether it has the shared write lock managed by the master process. If it does, it becomes the master process (204). If not, it determines from the shared control file which process holds the write lock (206) and it communicates to that process a request (208) to assume master process responsibilities. That process attempts to establish itself as master process (210). A test is performed (212) to determine if a new master process has been designated. If not, a race is conducted (214) between all shadow processes.
Abstract:
In a multi-system data sharing complex 10, 12, 13, a database system can write updated pages to a shared electronic store 16, for a fast write. Other database systems can obtain pages written to the shared store for further modification without the pages first being written to stable storage 14, 15. However, pages are eventually written to the stable storage in a castout process. Recovery of a database from failure of the shared store is bounded by determination of a recovery boundary which, when applied to the union of database system transaction logs, establishes a point in front of which are found log records of modifications to pages which were in the shared store when it failed. These log records are applied to page versions obtained from stable storage to recover from failure of the shared store.
Abstract:
Afin de verrouiller l'utilisation d'informations partagées dans un système multiprocesseur (100) ayant deux processeurs (101, 111) opérant de façon indépendante et asynchrone dont les unités de mémoire centrale (101, 112) reproduisent mutuellement leur contenu, un processeur doit provoquer l'exécution d'une opération atomique de lecture-modification-écriture (LME) par un sémaphore dans les unités de mémoire centrale reproduites des deux processeurs. Afin de régler correctement l'ordre d'exécution de telles opérations multiples de LME, des arbitres (106, 116) de bus des systèmes (105, 115) des deux processeurs sont mis en communication par un canal interarbitre (121). L'arbitre d'un processeur émetteur qui souhaite exécuter une opération LME le notifie à l'arbitre de l'autre processeur par le canal interarbitre. Des tentatives simultanées de notification par les deux arbitres sont tranchées en faveur de l'un d'eux désigné comme étant le maître. L'arbitre notificateur empêche son processeur d'exécuter une autre opération de LME jusqu'à ce que la première opération de LME ait été achevée, mais permet à d'autres opérations de se poursuivre normalement. L'arbitre ayant reçu la notification empêche son processeur d'exécuter une autre opération de LME jusqu'à ce que la première opération de LME ait été transferée par des connexions (107, 117) entre les processeurs et par un bus (120) du processeur émetteur au processeur de l'arbitre ayant reçu la notification, et y ait eté exécutée, mais permet à d'autres opérations de se poursuivre normalement. Des opérations multiples de LME sont ainsi exécutées par les deux processeurs dans le même ordre, de façon asynchrone et sans affecter l'exécution.
Abstract in simplified Chinese:内存子系统错误管理允许动态地改变锁步伙伴关系。内存子系统具有第一内存部分与第二内存部分之间的锁步伙伴关系,以使错误校正散布于内存资源之对。该锁步伙伴关系可被预先组配。回应于侦测到该锁步伙伴关系中之硬件错误,该内存子系统可将该第一内存部分与该第二内存部分之间的该锁步伙伴关系做取消或反向且创建或设置新锁步伙伴关系。该所侦测错误可为该锁步伙伴关系中之第二硬件错误。该内存子系统可创建作为锁步伙伴的第一内存部分与第三内存部分之间及作为锁步伙伴的该第二内存部分与第四内存部分之间的新锁步伙伴关系。该内存子系统亦可经组配来在改变合作时改变锁步伙伴关系之粒度。