34results about How to "Reduce context switching" patented technology

A prerequisite-based scheduler is disclosed which takes into account system resource prerequisites for execution. Tasks are only scheduled when they can successfully run to completion and therefore a task, once dispatched, is guaranteed not to become blocked. In a prerequisite table, tasks are identified horizontally, and resources needed for the tasks are identified vertically. At the bottom of the table is the system state, which represents the current state of all resources in the system. If a Boolean AND operation is applied to the task prerequisite row and the system state, and if the result is the same as the prerequisite row, then the task is dispatchable. In one embodiment of the present invention, the prerequisite based scheduler (dispatcher) walks through the prerequisite table from top to bottom until a task is found whose prerequisites are satisfied by the system state. Once found, this task is dispatched.

The invention discloses a user-mode file sending method, a file receiving method and a file sending and receiving device. The file sending method comprises the following steps: allocating a data reading and sending area, a data receiving and writing area and a data buffer area in the user-mode area; creating a data sending process; obtaining a file to be sent corresponding to the data sending process from the data buffer area; constructing metadata of the data sending process and recording the metadata to the data reading and sending area; sequentially processing one or more metadata of the reading and sending area, wherein the processing of the metadata comprises sending the file to a network adapter via a user-mode network device driver, and sending the file by the network adapter. According to the technical scheme of the invention, the data receiving and sending are completed by using the user-mode space of the data receiving and sending process, and the kernel state space of the process is not needed to completely shield the kernel, that is, the copy, the context switching, and the interruption can be remarkably reduced, and the performance of the data sending and sending is improved.

An apparatus, program product and method initiate, in connection with a context switch operation, a prefetch of data likely to be used by a thread prior to resuming execution of that thread. As a result, once it is known that a context switch will be performed to a particular thread, data may be prefetched on behalf of that thread so that when execution of the thread is resumed, more of the working state for the thread is likely to be cached, or at least in the process of being retrieved into cache memory, thus reducing cache-related performance penalties associated with context switching.

An apparatus, program product and method initiate, in connection with a context switch operation, a prefetch of data likely to be used by a thread prior to resuming execution of that thread. As a result, once it is known that a context switch will be performed to a particular thread, data may be prefetched on behalf of that thread so that when execution of the thread is resumed, more of the working state for the thread is likely to be cached, or at least in the process of being retrieved into cache memory, thus reducing cache-related performance penalties associated with context switching.

The invention discloses an operating system and a management method. The operating system comprises: a plurality of kernel service modules operated in kernel state corresponding to the system calling type, which are distributed on at least one processor and / or processor kernel; a plurality of application managing modules used for managing application procedure and application progress are distributed on different processor and / or processor kernel from the kernel service modules, which are used for managing application procedure and application progress; the kernel service modules are communicated with the application procedure through a system calling information. The management method comprises: step S1, the application system sends the system calling information, and then executes codeswithout dependent relation with the system calling result after sending the information; step S2, the kernel service module receives the system calling information and sends the system calling resultback to the application procedure; step S3, the application procedure receives the system calling result and executes codes with dependent relation with the system calling result.

The method for operating a file system comprises the steps of designing a virtual file to provide a result from the file directory for which a multitude of system calls is required, distinguishing the virtual file by a unique name from the real files of the file directory, and retrieving the result from the file directory by opening the virtual file and reading the content of the virtual file. The virtual file is designed in particular for a file system operation.

The invention discloses a memory migration method and a memory migration device. The memory migration method comprises the steps of: when a memory data migration request of a random source node in a storage system is received, dividing to-be-migrated data according to memory blocks on the basis of continuity of physical addresses of the to-be-migrated data, and determining the number of occupied memory blocks of the source node after dividing of the to-be-migrated data; acquiring memory blocks of which the number is same with the number of the occupied memory blocks of the source node after dividing of the to-be-migrated data in a target node in the storage system; and migrating the to-be-migrated data to the memory blocks in the target node according to the divided memory blocks. The invention further discloses a corresponding memory migration device. According to the memory migration method provided by the embodiment of the invention, through combining and dividing the to-be-migrated data according to the memory blocks, the memory blocks with continuous physical addresses are acquired from the target node; the to-be-migrated data are migrated from the source node to the memory blocks of the target node according to the memory blocks; and the number of context switching in migration can be reduced and a memory migration efficiency can be improved in the memory migration process.

The invention discloses a method for making a client operating system lightweight, a lightweight client operating system in a cloud environment, and a virtualized operating system based on the lightweight client operating system. The performance and functional limitations of the program in the cloud environment, simplify the design of the client operating system and improve its performance, and reduce the number of layers of abstraction and protection. The lightweight operating system mainly includes a virtual hardware driver module, a file system module, a network module, a memory management module, a thread scheduling module and a system call module. Among them, only one application program runs in a lightweight guest operating system, which reduces redundant and costly "isolation" in the guest operating system, uses a single address space, and uses one page table for all threads and kernels, reducing context switch. It can run c / c++ application programs and java programs, and can effectively improve the performance of application programs. It is an efficient application program running container.

The invention discloses a file sending method, a file receiving method, and a file sending and receiving device in a user state. The file sending method includes: allocating a data reading and sending area, a data receiving and writing area, and a data buffer area in the user state area. Create a data sending process; Obtain the file to be sent corresponding to the data sending process from the data buffer area; Build the metadata of the data sending process, and record it in the data reading and sending area ; Sequentially process one or more metadata in the reading and sending area, wherein the processing of the metadata includes: sending the file to the network card through the user mode network device driver, and the file is sent by the network card issue. According to the technical solution of the present invention, the user state space of the data sending and receiving process is used to complete the receiving and sending of data, without using the kernel state space of the process to completely shield the kernel, which can significantly reduce copying, context switching, and interruption. Improved data sending and receiving performance.