首页> 外国专利> METHOD FOR DEADLOCK RESOLUTION AND COMPUTER SYSTEM DEADLOCK REQUEST RESOLUTION

METHOD FOR DEADLOCK RESOLUTION AND COMPUTER SYSTEM DEADLOCK REQUEST RESOLUTION

机译:死锁解决方法和计算机系统死锁请求解决

摘要

One or more shared storage controller (SC) a plurality of central processor connected to the device (CPs) in the multiprocessor system consisting of input and output adapter requester (requester) between the potential deadlock (potential deadlock) the detection and to avoid discloses a hardware mechanism were (開 示). It needs to go to each memory controller may originate from outside sources, such as the internal source (source) as well as, CPs, input and output adapters, and other hardware equipment, such as the SC is used to store and fetch between the SC and the main memory processing. All requirements must be used to prioritize the flowchart is given priority before starting the run should be prioritized. Particular order of demand may lead to a deadlock (deadlock), require that the lock-out (locking out) or that has a lower priority because they use their priority cycle demand having a high priority, and any priority level, with the result required with the resources needed to complete the other requirements for. Herein in a deadlock resolution mechanism described in the present invention, checking the request register valid bit and uses the time pulses (timed pulse) a part of the pulse is used to detect the hang-up in the SC, the request is completed in the storage controller check the length of time spent in a valid state is not. A request register valid bit is turned on (on) for, when the two time pulse has been received is set in the internal latch hang-up detection. If at some point, whether the valid bit is reset, the detection logic and the internal hangup detection latch is reset. When the internal hangup detection latch set, the ongoing demand until the demand inside the hang-up is detected completion is allowed to be completed, a new request is pending in an inactive state (inactive state).
机译:一个或多个共享存储控制器(SC)连接到多处理器系统中的设备(CP)的多个中央处理器,由输入和输出适配器请求器(requester)之间的潜在死锁(potential deadlock)检测并避免公开硬件机制为(开示)。它需要去的每个内存控制器可能源自外部源,例如内部源(source)以及CP,输入和输出适配器以及其他硬件设备,例如SC用于存储和获取之间的信息。 SC和主存处理。在开始运行之前,必须先使用所有要求来确定流程图的优先级,然后再开始运行。需求的特定顺序可能导致死锁(死锁),要求锁定(锁定)或具有较低的优先级,因为它们使用具有高优先级和任何优先级的优先级循环需求,并且需要结果与完成其他要求所需的资源。这里,在本发明所述的死锁解决机制中,检查请求寄存器的有效位,并使用时间脉冲(定时脉冲)中的一部分脉冲来检测SC中的挂机,在该请求中完成存储控制器检查在有效状态所花费的时间长度不是。当内部锁存器挂起检测中已接收到两个时间脉冲时,请求寄存器有效位将被打开(接通)。如果在某个时候是否重置了有效位,则重置检测逻辑和内部挂起检测锁存器。当内部挂断检测锁存器置位时,允许正在进行的需求,直到检测到挂断内部的需求完成为止,新请求处于未激活状态(未激活状态)。

著录项

相似文献

  • 专利
  • 外文文献
  • 中文文献
获取专利

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号